Intel is about as ubiquitous as you can get in the enterprise technology landscape. Chances are, everybody on this planet who's interfaced with a computer has used the company's chips. That's a degree of universality enjoyed by only a handful of organizations throughout history. Yet, until very recently, the narrative around a once irreproachable institution has been dripping in corporate angst, with anxiety rising around a perceived decline.

Over the last few years, Intel's rivals have made up ground. We needn't dwell too much on Nvidia, which enjoys a de facto stranglehold on the US economy. Qualcomm is in a period of ascendency, pushing its chips into new kinds of devices like laptops. AMD is also enjoying something of a renaissance period, thanks in part to its data center dominance. Samsung has also been an ever-present in the chipmaking game.

Intel, however, has been bidding to "turn things around" for the best part of a decade. Plans have largely centered on manufacturing, including being a producer for "fabless" companies and building additional factories. Recent efforts include job cuts and reorganizations. Under fresh leadership in the form of Lip-Bu Tan, these efforts are now beginning to pay dividends, especially over the last six months. With graphics processing units (GPUs) now the hottest commodity in the AI build-out, can a pivot into this space be Intel's solution to finally exorcise its demons?

Intel's rise, fall and signs of resurgence – how did we get here?

Despite a positive history, events since the turn of the millennium have contributed to a perception that Intel has struggled to adapt to the modern world. To some extent, these views are grounded in reality – with material consequences to the company's bottom line. A sense of complacency and poor business decisions that have compounded over time, the narrative goes, has opened the door wide open to its much hungrier rivals, while executives have unfortunately been caught resting on their laurels.

"Turbulent is very appropriate," Alvin Nguyen, senior analyst at Forrester, tells ITPro, when asked how he would characterize Intel's performance as a company in the last few years. "I would say that until the past year, Intel had been struggling in terms of perception, thought leadership, and reputation."

It's easy to see why this is the dominant narrative, and the collapse in the company's stock price between 2021 and 2025 – a nearly four-times drop – illustrates how the market has perceived the company's performance. But how did we get here? One of the earliest missteps could perhaps be from the early 2000s, in a story now mythologized, that former CEO Paul Otellini turned down a deal with Apple to make iPhone processors.

You may also point to another former CEO Brian Krzanich, who in the mid-2010s held back on deploying next-generation equipment used in the extreme ultraviolet lithography (EUV) chipmaking technique in favor of multi-patterning technology. Commentators say this set the company back by half a decade.

Despite efforts to turn things around in the years that followed, including a $20 billion plan in 2021 to develop a foundry service, things did not improve, and the company continued its slide. However, following the retirement of Pat Gelsinger and appointment of its new CEO in March 2025, the company has undergone a significant internal transformation.

The firm made key executive hires in June last year, before axing 24,000 roles and cancelling factory plans the following month. Now, reports suggest that Intel will aim to buy into the success that some of its key rivals have lorded over it for years by investing in GPU production for AI data centers.

Can Intel tap into the GPU goldmine?

There's no overstating how important GPUs are in the AI buildout. The parallel processing capabilities of these chips, once used primarily in PC gaming, are primed to perform the sorts of operations and calculations essential to AI training and inference. Enterprise-grade GPUs are also highly sought after, with the likes of AMD and Nvidia dominating this space. So does the market need another player?

"Nvidia may be the leader, but they are unable to satisfy the market demand by themselves," says Nguyen. "As long as the demand for AI (and GPUs) keeps beyond capacity, Intel is not entering a "crowded" market. If there is a market correction or multiple market corrections, then this changes."

For what it's worth, Intel has been quietly improving the standard of its consumer-grade GPUs for years. The first generation of Intel Xe integrated graphics processors paved the way for a set of excellent entry-level GPUs in recent years in the form of the Alchemist and Battlemage Intel Arc Graphics GPUs you find in some of the best business notebooks.

The next generation of chips will be launched in the coming year or two under the Celestial banner, with Xe-3 integrated graphics serving entry-level machines. Nvidia, meanwhile, has been planning a scaleback in its consumer-oriented GPUs. Although the margins are less profitable than enterprise customers, is this something of a void that Intel can step into?

The next generation of chips will be launched in the coming year or two under the Celestial banner, with Xe-3 integrated graphics serving entry-level machines. Nvidia, meanwhile, has been planning a scaleback in its consumer-oriented GPUs. Although the margins are less profitable than enterprise customers, is this something of a void that Intel can step into?

"Intel is addressing both: AI-related enterprise demand provides an untapped revenue source, but the demands of AI are pulling in resources from other markets like consumer, that Intel also has an opportunity in the consumer space which they are targeting with consumer GPUs," says Nguyen.

The opportunity is certainly there. But, ultimately, it comes down to the execution. The early signs suggest that Tan is the man with the plan, with Nguyen telling ITPro that "Intel simply needs to keep doing what they are doing".

"They have focused on changing their corporate culture, became leaner, and became more customer focused," he explains. "They have had some clear wins in terms of their CPUs (specifically laptop CPUs) and announced big partnerships (US Government, Nvidia) that have changed perception about them [positively].

"Shipments and revenue were not as bad as the previous perception painted, so focusing on putting out products customers want, addressing issues instead of letting them stymie them for years, and making high-visibility partnerships that provide positive benefits will help grow positive perception."

Intel is on the right track, as far as Nguyen is concerned. Under Tan's leadership, the firm’s share price has surged from under $19 per share to highs of over $45. Although its previous turnaround efforts have been ineffective, seizing the opportunity to build industrial-grade GPUs might just be the boon that Intel needs to prove that its woes are firmly in the past and it can go toe-to-toe with the new kids on the block.

The US government is reportedly weighing up plans to take a stake in Intel as the chip giant’s troubles continue.

According to reports from Bloomberg, the proposed deal aims to support the company's efforts to expand manufacturing operations across the US.

A key focus here would reportedly be ramping up development of its delayed Ohio manufacturing facility. While initially touted to open this year, the company has repeatedly pushed back the launch date.

While first delayed until late 2026, in February this year Intel admitted the site wouldn’t become operational until 2030-2031.

The reports from Bloomberg come in the wake of sharp criticism from the US administration. Last week, President Donald Trump suggested CEO Lip-Bu Tan resign amid claims over alleged ties to China.

Subsequent talks between Tan and the administration appear to have been productive, with President Trump later hailing the meeting in a post on Truth Social.

"His success and rise is an amazing story," he wrote.

ITPro has approached Intel for comment, but received no response by time of publication.

In a statement given to Bloomberg, a spokesperson for Intel said the company is “deeply committed to supporting President Trump’s efforts to strengthen US technology and manufacturing leadership”.

“We look forward to continuing our work with the Trump Administration to advance these shared priorities, but we are not going to comment on rumors or speculation,” the spokesperson added.

Intel’s long-running woes

Intel has experienced significant troubles in recent years, falling behind competitors such as AMD and, notably, Nvidia, since the advent of the generative AI race. Intel's market value, for example, has halved since 2020.

The company has gone through repeated rounds of layoffs and a restructuring effort. Last month, the company announced plans to cut its global workforce by 15% and cancelled plans to develop new chip factories in Poland and Germany.

Tan’s appointment as CEO earlier this year was viewed at the time as an effort to steady the ship following Pat Gelsinger’s retirement. In the wake of the appointment, reports from Reuters detailed plans for an overhaul of the company’s chip and AI strategies in a bid to breathe new life into the company and reverse sluggish revenue growth.

Any involvement from the Trump administration could represent a lifeline for the semiconductor giant, and the White House has made clear its plans to ramp up domestic manufacturing and support for fellow chip giants in recent months.

Earlier this week, Nvidia and AMD announced an agreement with the administration to pay 15% of revenues from Chinese chip sales.

MORE FROM ITPRO

• Intel makes high-level hires while factory workers are warned of layoffs
• Intel layoffs confirmed as CEO eyes 'sustainable growth'
• Intel needs to “get its story right” to turn things around and capitalize on the AI boom

Welcome to ITPro's coverage of AMD Advancing AI 2025, coming to you live from the San Jose McEnery Convention Center. Hundreds of delegates are expected to fill the auditorium today to hear AMD CEO Lisa Su reveal the company's latest strategic plans for AI.

Su is expected to take to the stage at 9.30am PST and with her keynote slated to last for two hours, we can undoubtedly expect guests from both AMD and its partners to grace the stage.

With a little under two hours to go before Lisa Su's keynote starts, it's worth reflecting on what happened at the company's last Advancing AI conference, which only took place eight months ago in October 2024.

At that event, Su launched the enterprise-focused Ryzen AI Pro 300 Series CPU range, revealed the Pensando Selina data processing unit, and announced the arrival of Instinct MI325x accelerators. It's unlikely that the company has invited customers, partners, and press to hear updates on these products, but it could indicate the direction of travel.

At Computex in June 2024 Su announced the company was moving to an annual release cadence for its GPUs and with the launch of the Instinct MI325x accelerator series at the last Advancing AI we could see an update in this area.

The auditorium is starting to fill up at the San Jose McEnery Convention Center as we wait for Lisa Su to take to the stage for her keynote address this morning.

And we are underway with an opening video – apparently the key for the AI future is trust.

And here is the woman of the hour, Dr Lisa Su, CEO of AMD.

"I'm always incredibly proud to say that billions of people use AMD technology every day, whether you're talking about services like Microsoft Office 365 or Facebook or zoom or Netflix or Uber or Salesforce or SAP and many more you're running on AMD infrastructure," says Su. "In AI, the biggest cloud and AI companies are using instinct to power their latest models and new production workloads. And there's a ton of new innovation that's going on with the new AI startups."

As with many companies now – including partner Dell and rival Nvidia – AMD is talking up AI agents and its (positive) impact on hardware.

Su says that agentic AI represents "a new class of user".

"what we're actually seeing is we're adding the equivalent of billions of new virtual users to the global compute infrastructure. All of these agents are here to help us, and that requires lots of GPUs and lots of CPUs working together in an open ecosystem," she says.

AI isn't just a cloud or data center question now, it's also an endpoint question, and particularly the PC. "We expect to see AI deployed in every single device," she says.

Su hits on another theme that's becoming popular among many enterprise hardware providers: open source and openness.

"AMD is the only company committed to openness across hardware, software and solutions," Su claims. "The history of our industry shows us that time and time again, innovation truly takes off when things are open."

"Linux surpassed Unix as a data center operating system of choice when global collaboration was unlocked," she continues. "Android's open platform helped scale mobile computing to billions of users, and in each case, openness delivered more competition, faster innovation and eventually a better outcome for users. And that's why for us at AMD, and frankly, for us as an industry, openness shouldn't be just a buzzword. It's actually critical to how we accelerate the scale adoption and impact of AI over the coming years."

Su is now joined on stage by her first guest, Xia Sun of xAI, to talk about how the company behind Grok is working with AMD.

Sun focuses on how his small team is being bolstered by AMD hardware like MI300x and – of course – openness

Sun leaves the stage and now we have our first product announcement, the AMD Instinct MI350 Series.

"With the Mi 350 series, we're delivering the largest generational performance lead in the history of instinct, and we're already deep in development of Mi 400 for 2026," says Su.

"Today, I'm super excited to launch the Mi 350 series, our most advanced AI platform ever that delivers leadership performance across the most demanding models this series," she says, adding that while AMD will talk about the MI355 and the MI350 "they're actually the same silicon, but MI355 supports higher thermals and power envelopes so that we can even deliver more real world performance."

Su claims that the MI350 Series "delivers a massive 4x generation leap in AI compute to accelerate both training and inference". Some of the key claimed specs are:

• 288GB memory
• Running up to 20 billion paramaters on a single GPU
• Double the FP throughput compared to competitors
• 1.6x more memory than competitors

(These are AMD claims that haven't been independently verified by ITPro)

And now another guest: Yu Jin Song, Meta VP of engineering – Meta, of course, being the force behind Llama LLM.

"We're seeing critical advancements that started with [inferencing[ and now extending to all of your AI offering," says Yu Jin Song. "I think AMD and Meta have always been strongly aligned on ... neural net solutions."

"We try to shoot problems together, and we're able to deploy optimized systems at school. So our team really see AMD as a strategic and responsive partner and someone that we really like. Well, we love working with your engineering team. We know that we love the feedback, and you also hold the high standard. You are one of our earliest partners in AI," he adds.

Yu Jin Song explains that MI350 will become a key part of the company's AI infrastructure (it's already using MI300x accelerators).

"We're also quite excited about the capabilities of MI350x," he says. "It brings significantly more compute power and acceleration memory and support for FP4, FP6 all while maintaining the safety pull factor at mi 300 so we can deploy quickly."

These are some of the Instinct MI350 Series partners, which will have products coming to market with the chips in Q3 2025, including Oracle, Dell Technologies, HPE, Cisco, and Asus.

We're going from hardware to software, now, with Su saying this is "really [what] unlocks the full potential of AI". She leaves the stage and Vamsi Boppana, SVP of AI at AMD, takes the metaphorical mic to talk about ROCm.

"We have been relentlessly focused on what matters most, making it easy for developers to build better out of the box capabilities, easy setup, more collateral, stepping up community engagements," he says.

"We've been running hackathons, contests, meetups and mail, and our customers are deploying AI capabilities at unprecedented pace, and that's why we've significantly accelerated hour release cadence," he continues. "New features and optimizations are now shipping every two weeks. Leading models like llama and DeepSeek work on day zero."

He also points out AMD's partnerships with Hugging Face, Pytorch, OpenAI Triton, and Jax.

Now for his announcement: ROCm 7. This will, he says, "bringing exciting new capabilities to address ... emerging trends, and brings support for our MI350 series of GPUs".

"[It] introduces the latest algorithms, advanced features like distributed inference, support for large scale training and new capabilities that make it easy for enterprises to deploy AI effortlessly within ROCm 7," he says, adding that inference has been "the largest area of focus [and] we've innovated and invested at every layer of inference".

Boppana now invites his own guest to the stage, Eric Boyd CVP of AI platforms at Microsoft.

The two are discussing Microsoft and AMD's longstanding partnership, with Boyd saying Microsoft has been using "several generations of Instinct".

"It's been a key part of our inferencing platform, and we've integrated ROCm to our inferencing stack, making it really easy for us to take and deploy the models on the platform," he says.

Boppana is now talking about what he describes as a new opportunity for driving down inference costs: distributed inference.

"In any LLM serving application, there are two phases. There's the pre fill phase and there's decode phase," explains Boppana. "These two phases of the model are typically handled on the same GPU, but now if you [use] the same GPU, it often becomes a bottleneck for large models, or when demand spikes happen and you can get limited in performance or flexibility."

Disaggregating the pre fill and decode phase can "significantly improve throughput, reduce cost, and boost responsiveness", he says.

Here is that solution: ROCm Enterprise AI. "ROCm Enterprise AI makes it easy to deploy AI solutions with new cluster management software, it ensures reliable, scalable and efficient operation of AI cluster," says Boppana.

"Our mlops platform allows fine tuning and distillation of models with your own data and a growing catalog of models will come for specific industries," he adds.

Another announcement from Boppana: AMD Developer Cloud and Developer credits. All developers who are here will have a credit voucher in their email inbox right now, he says, which elicits cheers from the assembled crowd.

Boppana hands over to Anush Elangovan, CVP of software development, and Sharon Zhou, CVP of AI. Elangovan channels Steve Balmer: "Developers! Developers! Developers!" to yet more cheers – although he stopped after just three.

"We are serious about bringing rocket everywhere and to everyone, from client to the cloud," he continues. "In AI, speed is your mode. Access to compute is paramount. We've been delivering on speed, so now let's get you access to compute."

"With the developer cloud, anyone with a GitHub ID or an email address can get access to an instant GPU with just a few clicks," he says, directing developers who want to give it a go to dev.amd.com.

"We've also included a lot of easy to use frameworks like vlln, SG, Lang, pytorch, etc," he says. "You just select one of those frameworks, add your SSH key and then create and you're set."

Zhou briefly encourages developers in the crowd to "come say hi" at hackathons, university labs and elsewhere, before handing back to Boppana, who reveals the next of today's announcements: The ability to access ROCm 7 on Windows or Linux, on device or on client.

Boppana hands over to Forrest Norrod, EVP and GM of data center solutions, who is underlining the importance of agentic AI for AMD, as well as the performance improvements of fifth-generation Epyc CPUs.

He also talks up what AMD says is the continued importance of CPUs.

"Some will naively tell you that CPUs are less important in the age of AI, but that's not correct," he says. :With agentic AI, we see an explosion of autonomous agents accessing data and enterprise applications. This increases the needs for efficient, high performance x86 compute across the data center, then within the AI server itself, the CPU serves the demands of pre processing and workload orchestration to keep the GPUs working efficiently."

"Let me show you a few examples of how the right CPU can manage GPUs work better," he says. "Choosing poorly can create bottlenecks and stranded valuable resources. As you can see across a range of models and use cases, our fifth generation epic CPUs can boost the inference performance of the entire system from six to 17% that makes a huge impact on the overall TCO and performance of The AI deployment, and it's a critical element in designing the best possible AI system. So get much more out of your GPUs with the right CPU."

Lisa Su is back on stage to give the audience "a first look at the next big step for [AMD's] AI roadmap, the MI400 series". She quickly pivots from there to introduce Helios AI Rack.

"Think of Helios as really a rack that functions like a single, massive compute engine," Su says. " It connects up to 72 GPUs with 260 terabytes per second of scale up bandwidth. It enables 2.9 exaflops of FP four performance, and that is a great number."

Naturally, Su also takes a moment to make comparisons with AMD's competition, saying Helios supports "50% more HBM ... memory bandwidth, and scale out bandwidth".

And now, a surprise guest, "someone who's really an icon in AI", says Su: Sam Altman, CEO of OpenAI.

Altman talks about his company is seeing in the market.

"I think the models have gotten good enough that people have been able to build really great products, text, images, voice, all kinds of reasoning capabilities," he says.

"We've seen extremely quick adoption in the enterprise now, coding has been one area people talk a lot about, but I think what we're hearing again and again in all these different ways, is that these tools have gone from things that were, you know, fun and curious to truly useful," he says.

Asked by Su how compute demands are changing, Altman says: "One of the biggest differences has been, we've moved to these reasoning models. So we have these very long roll outs where a model that often think about come back with a better answer.

"This is really the pressure on model efficiency and long context for all this, we need tons of computers, tons of memory, tons of CPUs as well. And our infrastructure ramp over the last year and what we're looking over the next year has just been a crazy, crazy thing to watch."

Turning to the hardware used to support OpenAI's platforms like ChatGPT, Altman says the company is already running some work on the MI300x, "but we are extremely excited for the MI450, the memory architecture is great for inference. It can be an incredible option for training as well".

Su asks Altman what he sees in the future for AI. "How will the workloads evolve? What happens with – quote, unquote – AGI (artificial general intelligence)," she says.

"I think we're going to maintain the same rate of progress rated models for the second half of the decade as we did for the first," Altman responds. "I wasn't so sure about that a couple of years ago for a new research thing to figure out, but now it looks like we'll be able to deliver on that."

"So if you think forward to 2030 ... these systems will be capable of remarkable new stuff, about scientific discovery, running extremely complex functions throughout society, and things that we just couldn't even imagine," he continues. "It's really going to take these are huge systems now, very complex engineering projects, very complex research, and to keep on this [path of] scaling, we've got to work together across research, engineering, hardware, how we're going to deliver these systems and products.

"This has gotten quite complex, but if we can do on that, if we if we can deliver on that drives collaboration across the whole industry, we will keep this curve going," he concludes.

Altman leaves the stage and Su delivers some final thoughts to the audience.

"When I think about this past year, it's really redefined what progress in AI looks like. It's really moved at a pace unlike anything that we have seen in modern computing ... anything that we've seen in our careers, and, frankly, anything that we've seen in our lifetime," Su says.

"The future of AI is not going to be built by any one company or in a closed ecosystem. It's going to be shaped by open collaboration across the industry. It's going to be shaped because everyone is bringing their best ideas, and it's going to be shaped because we're innovating together. So on behalf of all of us at AMD, we look forward to changing the world with you together," Su concludes, and with that the keynote is over.

Global distribution giant TD Synnex has signed a deal to distribute the full range of Nvidia enterprise software and accelerated computing products in the UK. 

The new agreement means channel partners can now tackle their Nvidia product and platform requirements through TD Synnex, including Nvidia AI Enterprise and Nvidia Omniverse Enterprise software licenses for developers.

Businesses will also have access to the chipmaker’s high-performance networking adapters, accelerated computing GPUs, and superchips.

The distributor is also applying its accelerated delivery, flexible credit, and finance options, as well as its technical and professional services to the Nvidia range to help customers maximize their sales potential.

In an announcement, Simon Bennett, managing director of TD Synnex’s Advanced Solutions UK and Ireland business, said the company will provide the necessary support to help partners optimize Nvidia products within their infrastructures.

“Many organizations today depend on complex, hybrid infrastructures that encompass multiple cloud and on-premises resources,” he said.

“Nvidia technologies help drive the next-level graphics, HPC and AI performance essential for these intensive workloads.

“We will be working closely with Nvidia to make sure that our team is ready to provide partners with the technical and professional support and services they need to optimize Nvidia’s high-performance solutions, from the cloud to the data center and right out to the edge.”

A prominent global IT distributor and solutions integrator, TD Synnex serves over 150,000 customers across more than 100 countries around the world.

The firm’s portfolio pulls together IT products, services, and solutions from over 1,500 technology vendors, covering high-growth tech segments such as cloud, cyber security, big data and analytics, AI, IoT, mobility, and everything as a service.

With its new Nvidia distribution agreement, TD Synnex said it is expanding its expertise to meet UK partners’ full range of Nvidia-related needs, whether deploying solutions as part of their own designs or as a part of other systems’ brands.

“Nvidia technologies are used within the solutions from many leading vendors that we partner with, such as Dell, HPE and Lenovo, and now, with the extended distribution agreement, we can also provide the full Nvidia portfolio directly to partners as well,” Bennett added.

If you’re buying a new computer, there are two main CPU architectures to choose between. Windows PCs are normally built on the x86 platform, used by Intel and AMD, while Apple’s computers use the company’s own M1 and M2 processors, based on the ARM architecture.

There are differences between these approaches and significant implications for what it means for performance. 

ARM vs x86: Instruction sets

The x86 and ARM processor platforms do the same basic job, but they do it in quite different ways. Their internal logic is wired up in different arrangements, with different configurations of internal data registers and different sets of hard-coded instructions. On a fundamental level, they run software in different ways and use different code.

On the x86 platform, the internal structure and instruction set of the processor is ultimately based on that of the Intel 8008, an 8-bit CPU that debuted in 1972. In fact, machine code programs written for that chip can still be assembled and run on the latest processor from Intel or AMD.

Naturally, though, the hardware has evolved considerably since then. After the 8008 came the 8088, and then the 16-bit 8086, which powered the original IBM PC. In the 1980s this was followed by the 80186, 80286, and so forth – hence the “x86” moniker. 

Through generations, new features have been introduced to support multitasking and virtual memory; support has also been added for 32-bit and 64-bit operations, enabling computers to work efficiently with huge data sets and massive amounts of RAM. A series of extensions accelerate specific tasks such as graphics processing, virtualization, and data encryption.

Apple’s processors are based on the competing ARM architecture. This originated at Acorn Computers in the mid-1980s, at a time when the company was looking to create a successor to the hugely popular BBC Micro. Rather than buying chips from an external supplier, as it had with its previous home computers, the company set out to design a new processor that would outperform existing rivals. And it succeeded: at its launch, the ARM-based Acorn Archimedes was the most powerful home computer money could buy. 

Today the ARM platform is owned and developed by the Arm Group in Cambridge, and like x86, it’s continued to grow and develop since its inception. Successive versions of the platform have added 64-bit support and numerous extensions to speed up common mathematical operations – including, in the latest ARMv9 release, security, and artificial intelligence (AI) features. 

RISC vs CISC: The eternal rivalry

While ARM processors can do anything that x86 can, they have different strengths and weaknesses because they follow a different design philosophy, known as reduced instruction set computer (RISC). The name originally stood for “Acorn RISC Machines”, then later changed to “Advanced RISC Machines” as the market expanded beyond its original creator.

Its an idea that became popular in the 1980s and 1990s. This was a period when Intel and other chip makers were building more and more features and functions into the silicon, enabling programmers to execute complex operations with just a few lines of code. These processors came to be called complex instruction set computer (CISC) chips.

The RISC philosophy takes the opposite approach, aiming to make a CPU as simple as possible, by reducing it to a bare minimum of basic functions. Thus the ARM architecture uses just 34 instructions, which mostly handle simple mathematical operations and move data between registers and memory locations. By contrast, the Intel 8086 supported 81 instructions, permitting far more advanced data operations – and with subsequent revisions and extensions it’s ballooned to more than 200 instructions.

The RISC approach may seem counterintuitive. The smaller instruction set means that programs need to be longer and more complex to achieve the same results. However, a RISC chip can have a much simpler physical design than a CISC one. This can make it easier and cheaper to manufacture, and it can tear through instructions at a faster rate – in most cases, every operation is completed in a single clock cycle. It can consume less power too, which is why ARM processors are dominant in smartphones, where battery life is key.

While the CISC and RISC approaches are opposed, the differences aren’t as important as might be imagined. Few programs are written in pure assembly language these days, so developers don’t need to worry about the underlying architecture: they can write in Python, C#, or an alternative language before letting the interpreter or compiler deal with the translation. In fact, Apple’s ARM-based Macs include a real-time translation layer that lets them run programs written for x86 systems with no modifications.

The differences in power consumption are also smaller than they used to be. For many years, Intel struggled to match the low power consumption of ARM chips, not only because of the complexity of its CPU designs but also because its in-house manufacturing facilities were unable to reduce the size of the transistors inside its chips as fast as its rivals. That’s been a point of some embarrassment: the very latest Intel chips are still using a 10nm fabrication process (dubbed “Intel 7”), while Apple’s M-series processors have been using a 5nm process since their launch in 2020.

To help out, Intel’s 12th generation Alder Lake processors, released at the end of 2021, introduced a heterogeneous core design. Where previous Intel chips typically featured four or eight identical cores, current models combine lightweight ‘efficiency cores’ (E-cores) with powerful ‘performance cores’ (P-cores) that only roar into life when they’re needed for the most demanding tasks. This idea was actually pioneered by Arm – it introduced what it called the “big.LITTLE” design in 2011 – but now that Intel has got on board, we’re frequently seeing Windows laptops that can provide more than ten hours of video playback.

Which tech companies build processors?

Aside from the architectures, another notable difference between the two major computing architectures is that, unlike Intel, Arm doesn’t make any processors of its own. Rather, the company licenses its designs to companies that can then customize them as desired, and have them manufactured to their own specification. In the case of Apple Silicon, Apple uses the core ARM logic but adds many of its own optimizations, and outsources the manufacturing to TSMC.

The way chips are marketed to end users is different too. While all of Intel’s x86 processors use the same underlying architecture, it’s offered in an enormous number of different configurations. Within each generation of Core CPUs there are Core i3, i5, i7, and i9 variants, which further subdivide into ranges of different models aimed at mobile, desktop, or gaming systems. They all have different numbers of processing cores, different amounts of cache memory, different clock speeds, and different power requirements. It’s confusing, and when you’re choosing a computer there’s a risk that you could choose a model that’s underpowered for your needs.

Apple, by contrast, offers seven computer chips in total, namely the M1, M1 Pro, M1 Max, M1 Ultra, M2, M2 Pro, and M2 Max, at the time of writing. It’s a much simpler lineup than Intel’s, and even the regular M1 is competitive with a mid-range Intel chip.

How ARM and x86 CPUs access RAM

There’s one last difference between Apple’s chips and Intel’s – and this one isn’t intrinsic to the ARM architecture, but is a design decision Apple has taken itself. Where Intel’s chips rely on external system RAM, Apple incorporates the memory directly into the silicon die for its M-series processors. 

This means you can’t ever upgrade the memory on an Apple Silicon computer, which can lead to some agonizing decisions when it comes to choosing a specification. It also means that really large allocations of memory aren’t available at all on the mainstream chips: the M1 is offered with a maximum of 16GB of RAM, while the M2 is limited to 24GB. If you want 32GB or more you need to move up to an expensive M1 Pro, Max or Ultra system. For comparison, all of Intel’s 12th and 13th generation processors can use up to 128GB of RAM.

However, because Apple’s RAM is literally located right next to the processor logic, and connected to it via the fastest possible fabric, its processors can access code and data extremely quickly and efficiently. The standard M1 boasts a maximum memory bandwidth of 68GB/sec, while the M2 goes up to 100GB/sec, and the M1 Pro, Max, and Ultra models go up to 200GB/sec, 400GB/sec, and 800GB/sec respectively. With Intel, it all depends on the specifics of the processor, the RAM, and the motherboard, but even the newest, fastest Core i9 is limited to a theoretical maximum of 90GB/sec. 

What’s more, Apple uses what it calls a “unified memory architecture”, which means the whole range of memory can be directly accessed by either the CPU or the on-die GPU. This provides huge efficiency benefits compared to a conventional PC architecture, where the CPU and GPU each have separate memory banks, and can’t work together on the same data without copying it back and forth.

What makes AMD better than Intel?

There’s a third major player in the CPU market beyond Intel and Apple. AMD’s chips don’t have such a distinct identity, however, as they use the same core x86 architecture and instruction set as Intel.

Although AMD’s processors can run the same programs as Intel’s, there are some key differences. AMD sells its own chips, but it doesn’t manufacture them itself; this means it can use whichever foundry offers the best technology. While the first two generations of Ryzen CPUs were produced by Global Foundries, AMD switched to TSMC in 2019 to take advantage of its 7nm fabrication process, and the latest Ryzen 7000-series chips use the company’s 5nm process. That helps AMD chips spend more of their time running at the highest frequencies before they need to slow down and cool off.

AMD’s designs also frequently pack in more cores than similarly priced Intel chips, in part due to AMD’s ‘chiplet’ approach. Rather than build everything onto one die, it breaks the design down into multiple processor cores – chiplets – it then connects together, along with shared resources such as the main memory cache. Actual core counts can be misleading, as both companies use multithreading technologies that allow a single core to service two execution threads at once. Matters are further confused by Intel’s recent adoption of efficiency cores, which don’t contribute to peak performance.

Overall, though, you’ll normally get more multicore processing power from an AMD chip – and to support those cores, AMD tends to provide more on-chip memory than Intel. While Ryzen processors don’t put the whole RAM allocation on the chip die, as Apple’s chips do, they generally have large caches that help them keep processing data and instructions at full speed, without having to wait for information to be fetched from the DIMMs.

The only question is how valuable multicore performance really is. Big database servers and graphics rendering programs may benefit enormously from parallel processing power, but many desktop applications are mostly single-threaded. In practice, you might get a better experience with fewer, faster cores.

Last month, after a long delay, the UK government published its much-anticipated National semiconductor strategy. 

Originally due last year, the strategy’s announcement comes as the US-China chip war intensifies, with Taiwan – a key supplier of chips and home to TSMC – caught in the crossfire, raising concerns about supply chain security. 

The strategy’s publication also coincided with the G7 summit in Hiroshima, where the UK and Japan agreed a partnership to safeguard their respective chip industries.

“Semiconductors underpin the devices we use every day and will be crucial to advancing the technologies of tomorrow,” said prime minister Rish Suank in a statement. “Our new strategy focuses our efforts on where our strengths lie, in areas like research and design, so we can build our competitive edge on the global stage.”

Up to £1 billion of subsidies will be made available through the strategy over the next decade, with an initial £200 million to be spent between 2023 and 2025. A significant portion will be directed toward expanding the compound open foundry ecosystem. Compound semiconductors have capabilities beyond silicon chips found in smartphones, and are used in electric vehicles, renewable energy technologies and satellites. 

Why the UK is targeting compound semiconductors 

Compound semiconductors are a considerable growth opportunity for the UK, says Mark Dickinson, CEO of London-based Intrinsic Semiconductor Technologies. In March, the company announced it had raised £7 million in funding to develop a resistive random-access memory (RAM) technology that addresses the memory bottleneck caused by current external flash memory solutions. 

Dickinson explains the UK can’t manufacture silicon chips without the help of other nations. Yet, becoming a leader in compound semiconductors, for example, “would allow us to form reciprocal supply arrangements for silicon chips with potential partners”.

Mark Lippett, CEO of Bristol-based XMOS, which designs fabless chips for embedded systems, agrees with this sentiment. XMOS is mentioned by name in the UK’s strategy.

“The UK won't be able to develop a full supply chain to fulfil its semiconductor needs – this is beyond the scope of even the EU or US. The question is, how does the UK secure a seat at the table. And, the answer is by having technology that others want,” argues Lippett. 

An example of this in action is the Netherlands, whose golden goose, ASML, is a leader in the photolithography machines market. Photolithography uses light to etch integrated circuits onto silicon wafers – ASML’s machines are wanted by chipmakers around the world and its customers include Intel, Samsung and TSMC. 

An underwhelming level of investment 

While the UK has finally joined the subsidies race, the level of investment is smaller than the £2.5 billion the government has committed to quantum computing, while it also pales in comparison with packages offered by the EU and US.

The former agreed a deal in April to spend €43 billion (£37 billion) to try to double its share of the global semiconductor market to 20% by 2030. The US pledged $52 billion (£42 billion) to its chip industry to counter China as part of the $280 billion CHIPS and Science Act signed by President Biden last August. 

Dickinson is disappointed at the low level of investment pledged by the UK and thinks “there is no way we can catch up” to the EU and US. 

“It leaves the UK in a very vulnerable position – both in terms of security of supply and advancement in research and development. We will be left behind in this critical sector of the economy,” adds Dickinson.

The government is hoping to mitigate any problems by leveraging existing expertise in tech clusters, including Cambridgehshire and South Wales. Words such as ‘skills’, ‘training’ and ‘talent’ appear dozens of times in the strategy, a sign that the government recognises the importance of nurturing and developing home-grown talent that will help design and build next-generation chips.

One thing the strategy doesn’t mention, though, is how overseas talent can be used to plug the short-term skills gap. 

Supporting the startup ecosystem 

The success of the strategy is ultimately going to hinge on the support the UK’s chip startups receive to launch and grow. To this end, the Department for Science, Innovation and Technology (SIT) is expected to review pre-seed and seed funding opportunities made available through incubators. 

“We need more detail on the specifics,” warns Scott White, founder of Cambridge-based Pragmatic Semiconductor, another company that is mentioned in the strategy. “Scale-ups today face evermore enticing opportunities from abroad and the government needs to address the very prominent later stage funding gap in the UK – and quickly.”

Ekaterina Almasque, general partner at deep tech venture capital firm OpenOcean, believes the government has to have a more strategic vision. 

“Funding pledges are important to get the ball rolling in driving innovation and growth, but alone they are not enough,” says Almasque, who led early investment in Bristol-based Graphcore, whose chips are used for artificial intelligence applications and is also mentioned in the strategy. 

The UK needs to ensure that it’s investing in the data economy and supporting the startups developing data-intensive applications that will help to fuel demand for semiconductors and power the future businesses, governments and academia. 

“This will enable domestic companies to scale and compete with international rivals,” adds Almasque.

If the UK’s semiconductor startups and scale-ups are held back by what they perceive to be a lack of support, they could potentially relocate their operations and manufacturing overseas. They’d be far more likely to thrive away from the UK, and could get their hands on a bigger slice of a bigger subsidy pie. 

If this happens, it could also trigger an exodus of innovative companies, especially those that want to shorten their supply chains. Homegrown talent might then follow, resulting in a brain drain and pushing the UK down the global semiconductor pecking order. 

Revenue in the global semiconductor industry is set to plummet by more than 11% over the next year as waning business demand for hardware impacts manufacturers. 

An analysis from Gartner this week projected an 11.2% decline in revenue to around $532 billion, marking a significant decrease from a market total of $599.6 billion in the year prior.

A key factor in the predicted revenue contraction is the consistent decrease in demand witnessed across the PC, tablet, and smartphone semiconductor markets, which Gartner warned is ‘stagnating’. 

Tightened budgets at businesses worldwide mean that semiconductor manufacturers will be faced with an increasingly competitive and strained marketplace. 

These combined markets will account for around 31% of all semiconductor revenue in 2023, totalling $167.6 billion. 

“As economic headwinds persist, weak end-market electronics demand is spreading from consumers to businesses, creating an uncertain investment environment,” said Richard Gordon, practice VP at Gartner. 

Gordon added that a rampant oversupply of semiconductors will also have a significant impact on the market. 

This saturation of chips is “elevating inventories and reducing chip prices” across the board, which in turn will accelerate the decline of the market across 2023. 

Tough conditions for hardware demand

Demand for IT hardware has been consistently decreasing in recent months due to deteriorating economic conditions. 

Research from Gartner in January revealed that global PC shipments declined to record levels in the final quarter of 2022, dipping to 65.3 million and representing an overall decline of 28.5%. 

Budgetary constraints among global businesses, combined with a growing trend of firms delaying purchases and extending lifecycles were all highlighted as key contributing factors, Gartner said at the time. 

This trend has continued into early 2023, analysis shows. Earlier this month, IDC revealed that global PC shipments dipped by nearly one-third in the first quarter of 2023, highlighting this continued decline. 

The research from IDC showed that total shipments of PCs declined 29% in Q1 to 56.9 million units. IDC said that weak demand, excess inventory, and a “worsening macroeconomic climate” were responsible for the drop in shipments.

If you’ve tried to buy anything with a semiconductor inside it over the past few years, you’ll know that it has been a nightmare. Supply chain problems and shortages since the pandemic have driven many companies and customers to despair.

Many had, however, hoped the crisis would be past its worst by the end of 2022. And indeed, by the end of last year, there was a glint of light at the end of the tunnel. “Supply chain update – it’s good news!” wrote Raspberry Pi founder and CEO, Eben Upton, on the company’s blog. In the post, he described a much rosier picture for 2023 supplies of the company’s wildly popular single-board computers. So can we look forward to supplies returning to normal?

Unearthing hidden inventory pools

The initial supply shock was one that almost every hardware firm experienced. “We had finished goods inventory, we had component inventory, and we had a pipeline,” Upton explains. “We were probably good through until the second quarter of 2021.”

But then problems began to emerge further down the supply chain. “What you started to see was lead time blowouts on components,” he says, adding that component deliveries that were typically expected to take 26 weeks suddenly became much longer. He recalls how some vendors, such as the manufacturers of microcontrollers, would suddenly quote him a two-year wait for parts.

“A big lesson for us was that a 104-week lead time doesn’t mean you can have your chips in 104 weeks, it means ‘go away’,” he says with a chuckle.

The trick then was to carefully manage what stock was available, to make the most of the limited supply. “The thing that sustained us through 2021 was having component and finished goods inventory pools in our organisation,” says Upton. “The thing that sustained us last year was going further down the OEM supply chain, and trying to detect where the hidden inventory pools there were, and then trying to thin them out.”

This involved some awkward phone calls. “If someone calls up and says ‘we want 10,000 Raspberry Pis’, we say ‘Really, though?’. Are you really going to use them or are you building inventory?,” says Upton, adding that he paid particular attention to OEMs who build products with his Raspberry Pi computers inside. He didn’t want customers to buy them and just leave them sitting on a shelf. “If we get a chip from a vendor, we want to turn it into a Raspberry Pi as quickly as possible. And then we want the OEM to turn it into a finished product as quickly as possible and sell it to their customers as possible.”

This wasn’t just about Raspberry Pi’s business either – as Upton saw the impact supply chain issues could have downstream. “Most of our OEM customers [buy] 100 units a quarter, 100 units a year,” says Upton. “These are little companies that are making little widgets with Raspberry Pis. And if they don’t get the Raspberry Pis, they’re going out of business.”

He explains how the company could have mitigated the supply problem in two ways: the “American” solution, of putting up prices until demand falls or the “Soviet” solution of rationing. He chose the latter. “If someone designs my product into their product, we don’t put the price up on them after they’ve committed to us. They’ve placed their faith in us,” says Upton. “You could use price as a mechanism to control demand. But it just didn’t feel like it didn’t feel like the moral thing to do.”

Clinging onto reasons for optimism

The good news is the future is looking more promising as the world economy shakes itself back to normal after the shock of COVID-19. “This isn’t that supply went down, it’s demand went up and supply couldn’t flex up to meet it, for a variety of reasons, mostly associated with the amount of time it takes to build new facilities,” says Upton. “For silicon vendors and for us, this has been about smart demand management. People have succeeded or failed based on how hard they’ve worked.”

So, supplies are improving and Upton expects the flow of new Raspberry Pi devices to be back to normal by later in the year. “You’ll see improved availability for a variety of products going through the year for hobbyists, I think you’ll see single unit [sales] limits, at least till the end of Q2, probably into Q3. But you’ll see allocation progressively lighten up on more and more products as chips flow through. And then Q3 onwards, we’ll be back [to normal].”

According to Upton, despite all of the problems, the company still managed to ship seven million units in 2021 – matching the 2020 total. But without the supply chain improving, it might not have been possible to repeat the trick for a third year running.

“I’m glad it’s coming to an end because I think we’ve learnt lots of cool tricks, but a lot of the cool tricks are single-use tricks,” he says of the company’s supply chain management. “Once you’ve detected a pool of inventory in the system, you can’t deplete it again. and the problem is our bag of tricks is not very full anymore.”

The forthcoming adoption of USB-C as the connector of choice has been a frequently covered topic and, as you will have read, it’s been causing quite a heated discussion. Since 2009, there’s been a voluntary code in place between the EU and mobile manufacturers to reduce the variations of chargers from over 30 down to three. Last September, the EU Commission published an agreement, bringing this to one.

The rule change is being championed for both consumer convenience and to help fight e-waste. According to EU figures, consumers spend €2.4 billion on standalone chargers and over 11,000 tonnes of them end up in e-waste streams. The hope is that changing to a single socket will make it simpler to find something suitable to stick in it, but also control how rapidly your device will charge. Fast-charging technologies are also part of the rules.

Despite this genuinely positive news, what has been omitted from the legislation – in a cruelly deliberate oversight – is the effect fixing USB-C ports has on ageing Yorkshire-based PC technicians whose eyesight has lost the sharpness of its youth. A typical USB-C port has 24 internal pins, which obviously need to be connected to the machine. Comparatively, a USB-A port has up to five internal pins and when you consider the size differential between the two sockets, soldering up to 24 pins within a smaller area increases the difficulty level somewhat.

Remember, USB-C purely describes the shape of a port and not its functionality, but because it’s so versatile, USB-C is currently the obvious candidate for universality, irrespective of how much of a nightmare it is to fix. Looking ahead, the USB-C legislation won’t encapsulate laptops until 2026, but as the writing is on the wall, manufacturers could move early, triggering some radical redesigns inside the machines.

Traditionally, fixed ports such as USB, DisplayPort, Ethernet or HDMI are always soldered to a PCB, but most DC power barrels have broken free in recent years and are connected with a simple plug and socket. This makes repair much easier for anyone confident enough to open their laptop, or cheaper if professional help is required. Broken sockets remain the number one hardware fault I encounter on laptops, so when USB-C becomes the de facto method of power delivery, it’s critical that swapping a broken USB-DC is at least as simple as the plug and socket we have now. Keeping it on the board vastly increases the repair costs and the probability of the machine going in a skip.

It’s taken a long time to get this far, but now universality has become mandatory, it should be simpler to evolve the rules in the future. Jump forward five years when we’re talking about 'USB-D', the established rules can be tweaked. It’s not perfect but it draws a line in the sand and is far better than an office full of workers using 30 different chargers.

For decades, Intel’s relationship with chipmaking rivals such as AMD and Nvidia has been a feisty one, with the industry giant often seeking to exploit its domination at all costs.

In 1999, for example, the company got into a tangle with the US Federal Trade Commission (FTC) over how it was protecting technical information and patents. A decade later, the European Commission found “abuse of dominance” as Intel paid retailers and computer manufacturers to only use Intel CPUs.

The tables have turned in recent, years, however, with Nvidia becoming the most valuable US chipmaker in July 2020, and the company forced into a $20 billion plan to revitalise its business the following year. Part of this transformation has been a pivot towards acknowledging the strengths of its competition while venturing further into the cloud and software spheres.

Nevertheless, given its history, it’s perhaps a little unexpected that Intel, given its investments in the SYCL programming language, is now effectively offering an olive branch of friendship and asking its rivals: can’t we all just get along?

Speaking a universal language

“We should have an open programming language… that can run on Intel, AMD and Nvidia GPUs without changing your code,” Intel CTO Greg Lavender told the business site Venturebeat last October, as he sought to promote the company’s investments in SYCL, a Java-like programming language that bridges the gap between different CPU and GPU hardware.

It’s a language designed for software where processes are “heterogeneous” – tasks such as processing large amounts of data in parallel where you might throw a mass of GPUs at the problem. As such, it’s already used by various scientific supercomputers and in software such as GROMACS, which scientists use to simulate the behaviour of proteins, and which came in very handy during the early days of the pandemic.

What makes SYCL a significant break for Intel is that it is entirely open, and is by definition designed to make the same code work across different GPUs and CPUs. In theory, you could write the code and have it run on everything from an Intel i3 to an AMD Ryzen or an Nvidia GeForce GTX.

To really show it meant business, Intel last year bought Codeplay, a company that has contributed extensively to SYCL, as part of what it calls its “open standards vision”.

Making code cheaper

“Writing software is really expensive,” says Dr Tom Deakin, a lecturer in advanced computer systems at the University of Bristol, and chair of the SYCL advisory panel, who argues that given the choice, people want to write portable code that will run on different hardware.

“If they're going to have that code live for a long time, they can't afford to keep rewriting it.”

SYCL is not the first programming language to attempt to bridge the gap between different types of hardware, with the likes of OpenCL (which is managed by Kronos, the same group that runs SYCL), but what makes it different is the level of abstraction – essentially taking care of high-level processes, making it quicker and easier to write code.

Nvidia supports a similar toolkit called CUDA, but because it’s proprietary, it only works on Nvidia chips, which is perhaps why Intel is so keen to grow an alternative that can work on its chips too. “People don't really want to write CUDA to do that, because it locks them into only being able to run their programmes on Nvidia and not any other system,” he adds.

He argues this is good for coders, as it means there’s no risk of relying on the fortunes and decisions of only one company. “There's lots of compiler support for it and lots of those compilers support multiple devices from multiple vendors,” says Deakin. “I always like to think this multiple compiler position is the sign of a healthy ecosystem.”

Cracking the GPU market

This perhaps reveals why pushing SYCL isn’t an entirely altruistic move from Intel. The company has a close to 70% share of the GPU market if you include integrated graphics chips on devices such as laptops, but it controls only a tiny proportion of the discrete GPU market, which is dominated by Nvidia. SYCL could be an attempt to challenge Nvidia’s domination of GPUs.

“The choice they [Intel] had was do their own thing with their own badge, or try and build on something that's a bit more open,” Deakin continues. “If they made their own CUDA, then they'd make something that wouldn't be portable, and it's not attractive to break into that market.”

The move also aligns with Intel’s broader strategic shifts. The firm is in the process of reorganising its business away from controlling the entire chip ecosystem, to a more mercenary approach, where it will build whatever chips customers want.

“Their choice was an open approach and I think that’s a good thing,” says Deakin. “It means [developers] can still run programmes on Nvidia if they want to do that, but they can also run on Intel and it opens up the opportunity to compete.”

While it's easy to be swept away by Intel's Core i9 chips and high-power computers such as the new MacBook Pro, there are some tasks that simply don't need that much firepower. Especially if the PCs are going to be left on 24 hours a day, seven days a week, chugging through undemanding tasks.

That's the central idea behind Intel's NUC 11 Essential mini PCs. They're designed for scenarios where you switch the device on and then forget about it, perhaps for years.

Think digital signage, media streaming, and shared storage. Start totting up the numbers and power consumption becomes far more important than how many high-performance cores are on hand.

Intel sent us two configurations of its NUC 11 Essential barebones PC that exemplify this very different kind of computer. All you need to add is RAM – two SODIMM sockets lie empty – and an M.2 2280 SSD. Load your operating system of choice and you're away.

The cheaper version of the NUC 11 Essential comes with a two-core Celeron N4505 inside, the glamour model with a four-core Intel Pentium N6005 chip. It isn't too great a simplification to say that the Pentium version is twice as fast. In the multicore sections of the Geekbench 5 test, their scores were 1,201 and 2,385 respectively, and that pattern repeated in Cinebench R23 with results of 1,140 versus 2,381. Likewise in the IT Pro benchmark, with 45 against 83. To put those scores in perspective, we used to describe laptops that scored around 80 as fast, but would rarely recommend a system that scored less than 50.

We saw the same pattern in 3DMark Time Spy, with 207 versus 459, but both of those are woeful results that indicate the NUC 11 Essential isn't destined for Minecraft, let alone Cyberpunk 2077.

Due to the way PCMark 10 works – it doesn't push machines to extremes, just sees how they handle a variety of tasks – the gap between the machines in that benchmark is much narrower, but the 660 difference you see in the graphs below counts for a lot in the real world. If we only needed a PC for web browsing, email and sending rude messages to friends on Twitter, then the Pentium system is more than fast enough. The Celeron version, however, would drive us crazy due to its stuttery nature.

However, if you're going to turn this machine into a firewall or a kiosk PC – any usage where it simply chugs along rather than one where the processor must meet spikes of multitasking demand – then the Celeron version saves you cash. Potentially in the long run, too: both machines sip 6W of power when running idle, but the extra cores in the Pentium chip meant its peak power demands rose to 32W compared to 14W for its two-core sibling.

The units themselves are identical. To describe them as small, square slabs of plastic is both true and unfair; it makes them sound cheap and wobbly, when in reality these are well-made shells that will probably look just as good in ten years time as the day you take them out of the box. Intel backs them with a reassuring three-year warranty, too. But they are indisputably tiny: you could almost squeeze four NUC 11 Essentials onto a page of IT Pro.

Intel makes it easy to access the insides, with four crosshead screws standing in your way, but there isn't much to do once you get there. You can replace the Wi-Fi 5 M.2 adapter easily but, aside from the aforementioned M.2 SSD slot and SODIMM sockets, that's it for upgrades. And the NUC 11 Essential only supports Gen3 SSDs, not the faster Gen4 drives.

Nor is this barebones PC designed for fast external storage. There's no USB-C here, and the two front USB-A ports only support USB 3.2 Gen 1 (so 5Gbits/sec). There's no integrated mic, so if you want to make calls then you'll need to use Bluetooth or the front-mounted 3.5mm jacks. Also note the Kensington lock on the right-hand side.

There are many more connectors at the rear, including the DC input for the separate power supply. You'll find two USB-A 3.2 Gen 2 (10Gbits/sec) ports along with twin USB-A 2 ports, an HDMI 2.0b output, DisplayPort 1.4 (you can attach two monitors at the same time) and a gigabit Ethernet port. A sensible selection for, say, a basic web server or media centre.

Or even for desktop PCs for office staff. Not the Celeron version – you'd have a riot on your hands – but buy the Pentium edition of the NUC 11 Essential, then add two 8GB SODIMMs and a cheap SSD, and you have a perfectly serviceable Windows 11 computer for under £300. Although that doesn't include the cost of the operating system.

A fleet of these mini-systems idling at 6W rather than a typical PC's 50W could make a huge difference to your monthly running costs. And if something does go wrong with one of them, it won't cost much to have a few spares in stock.

Intel NUC 11 Essential specifications

The T2 security chip is a specialised component developed by Apple and used in Intel-powered macOS machines. It’s designed to secure the boot process and provide hardware-level encryption for data stored on the machine, among other features, offering an additional layer of security for enterprise users.

One of the most significant features of the T2 chip is the inclusion of a Secure Enclave coprocessor – used to power Touch ID and handle other secure operations. It operators much in the same way as the Trusted Platform Module (TPM) chip in Windows devices.

A successor to the T1 chip, introduced in 2016, the T2 chip was fitted into Mac devices from the following year, although recent machines powered by Arm-based CPUs, including the 13in M1 MacBook Pro (2020) lack this component. This is because the T2 chip’s security features are directly built into Apple’s new line of Arm-based processors. M1 devices, and more recent machines, have their own Security Enclave that allows a single chip to manage all the security features the T2 chip would traditionally oversee.

When did apple launch the T2 security chip?

Apple first introduced its T2 security chip in 2017 with the iMac Pro, as a custom version of the A10 chip, which was previously used in the iPhone 7 and iPad Pro. The T2 chip was later included in other Mac models, starting with the 2018 MacBook Pro.

Apple devised the T2 chip as a way to bolster the security of its macOS machines. It was designed to provide additional security features such as a secure boot process, hardware-level data encryption, and Touch ID, as well as to enable features like Activation Lock.

Prior to the T2 chip, Mac computers relied on software-based security measures and lacked dedicated hardware for handling secure operations. The T2 chip was designed to provide an additional layer of hardware-based security for Macs. helping to protect against cyber security threats such as malware.

How does the T2 chip work?

The Apple T2 security chip is a specialised component and includes a Secure Enclave coprocessor used to handle secure operations on the Mac. It sits independent to the Intel processor and the operating system.

The T2 chip is responsible for managing the boot process on the Mac and ensuring that only trusted software is allowed to run. It does this by checking the integrity of the boot process and verifying that all software being loaded is signed with a trusted certificate. If any untrusted software is detected, the boot process is stopped to prevent it from running.

The T2 chip is also used to provide hardware-level data encryption for the Mac. It includes a hardware-accelerated AES engine, which is used to encrypt and decrypt data stored on the machines’s internal storage. This provides an additional layer of protection for sensitive data and helps to prevent unauthorised access to the data.

The Secure Enclave coprocessor within the T2 chip is used to power the Touch ID fingerprint sensor and handle other secure operations on the Mac. It’s responsible for storing and protecting the user's fingerprint data and for handling secure transactions such as Apple Pay payments. Overall, the architecture of the T2 security chip is designed to provide additional security features in such a way that just a standard Intel CPU cannot.

What are the benefits of a T2 security chip?

The T2 chip provides additional security features, such as a secure boot process and hardware-level data encryption, which can help to protect against various threats. The addition of a T2 security chip to Intel-powered Macs offer an added layer of assurance, particularly for users handling sensitive data and those who need their machines for sensitive operations.

The security coprocessor built on the foundations of the T1 chip to offer secure boot and real-time encryption and decryption functionality. Alongside added convenience of securing biometric login, and using features like Apple Pay, the T2 chip’s hardware-accelerated AES engine improves performance of encryption and decryption. There are a number of improved and entirely new features that contribute to these benefits.

Secure boot process

The T2 chip checks the integrity of the boot process to ensure that the Mac is running only trusted software. The T2 chip's secure boot process helps to protect against threats such as malware and can help to prevent unauthorised access to the device. This can be important for enterprises that handle sensitive data and need to ensure that their devices are secure.

Hardware-level data encryption

The T2 chip can encrypt data stored on the machines internal storage, providing an additional layer of protection for sensitive data. This can be especially useful for business users and enterprises who may handle sensitive data and need to ensure that it is protected from unauthorised access.

Powering Touch ID

With its Security Enclave, the T2 chip powers the Touch ID fingerprint sensor, allowing users to unlock their Mac and make purchases with Apple Pay using biometrics. This can be convenient for enterprise users who need to log in to their Mac frequently and may not want to enter a password each time.

Activation Lock

The T2 chip helps to prevent unauthorised access to a Mac that has been lost or stolen by enabling Activation Lock, which requires the user's Apple ID and password to be entered before the Mac can be used. This can be useful for businesses that need to protect their data and prevent unauthorised access to company devices.

The T2 security chip's features can provide additional security and convenience for enterprises, helping to protect against threats such as malware and unauthorised access to the device, and providing a convenient way for users to log in to their Mac.

What are the concerns around the T2 security chip?

While the T2 security chip provides additional security features for Macs, it isn’t without its issues and concerns. It is crucial for users to be aware of these potential issues and to stay up to date with software updates to address any vulnerabilities.

Repairability

Some users and repair technicians have raised concerns about the T2 chip's impact on repairability. Because the T2 chip handles many security-related functions on the Mac, replacing it can be difficult and may require specialised equipment. This has led to concerns that the T2 chip may make it more difficult and expensive to repair certain models.

Security vulnerabilities

Like any software or hardware, the T2 chip is not immune to security vulnerabilities. In 2020, a security researcher discovered a vulnerability in the T2 chip that could allow an attacker to gain access to the Secure Enclave coprocessor, potentially exposing sensitive data stored on the Mac. Apple released a patch to fix this vulnerability.

User privacy

Some users have raised concerns about the T2 chip's ability to record microphone audio even when the Mac is in sleep mode. While Apple has stated that this feature is intended to enable Siri to respond to voice commands while the Mac is asleep, some users have raised concerns about potential privacy implications.

When Raspberry Pi went on general sale ten years ago, the bare bones, low-cost computer harked back to the past while looking towards the future. The idea was that large numbers of users would learn how to program computers once more, evoking memories of the era of the BBC Micro, ZX Spectrum and a myriad other machines on which millions of people cut their coding teeth.

But while many Raspberry Pi users have done just that, the computer has proven fruitful in other ways. The various versions of the small single-board computers have formed the heart of many projects, from magic mirrors and weather stations to robots and orreries. But an ability to emulate vintage computers and consoles has arguably caught users' imaginations the most.

The free software library RetroPie, for example, lets people relive the Amstrad CPC, Dragon 32, Commodore 64, Master System, Neo Geo, Oric, TRS-80 and Vectrex among many, many other machines. Raspberry Pi has also effectively become a near-perfect Amiga thanks to the distro Amibian.

Recently, however, our heads were turned by a new tool created by Jarosław “Jaromaz” Mazurkiewicz called MacintoshPi. This open-source project allows users to run full-screen versions of Apple's Mac OS 7, Mac OS 8 and Mac OS 9 (operating systems released between 1991 and 2001) and it does so complete with sound, an active internet connection and modem emulation.

A growing love

Jaromaz is an IT team leader for EuroLinux, a company building a Polish clone of the Red Hat Enterprise Linux distribution. Hailing from Poland, he first came across Apple's computers in 1998. By that time, the company had only been selling its wares in the country for seven years. “Before 1992, Eastern Bloc laws restricted the flow of Western technology so, during my childhood, we only had access to ZX81, ZX Spectrum, Atari and Commodore computers,” he tells IT Pro’s sister publication, PC Pro.

Although some Polish schools snapped up Macintosh Classic computers when they became available, Windows PCs dominated. “My youthful memories are associated with Windows 3.1/95/98/ME and Red Hat Linux,” says Jaromaz.

Yet when working in the Polish editorial office of PC Magazine, he was able to see Macintosh computers close up. “Mac OS systems were being used for the typesetting of newspaper titles and I used to watch these people working,” he says. “Sometimes they provided me with Apple computers for testing.”

It was a while before he reacquainted himself with Macs, however. “In 2015, I built a hackintosh with OS X Yosemite for testing purposes because I'd seen the Retina screens in Polish Apple Reseller showrooms and the resolution and readability of the fonts fascinated me,” Jaromaz explains.

“The hackintosh convinced me of the Apple system itself – its attention to detail, cleanliness and visual perfection. Shortly after that, I purchased my first MacBook Pro with Retina screen and I also installed the Basilisk II emulator to begin serious, long tests of Mac OS 7.5.5.

“I was surprised at how many programs and games worked seamlessly with the emulator and I knew many of them from the PC I owned in my youth, although some were brand new to me. This gave me the chance to explore the most interesting Mac OS titles retroactively.”

It was then that he fell in love with Mac OS 7.5.5. “It's so friendly, intuitive, elegant and easy to use,” Jaromaz says. He dreamed of running the operating system with the original Macintosh Classic case.

“But as a software, not hardware specialist, I didn't want to buy older and expensive hardware only to worry that it would have a short lifespan,” he explains. “I believed a system emulated on a modern ARM board would successfully replace the original hardware and create a solution.”

Introducing MacintoshPi

At first, Jaromaz looked to run the early Macintosh emulator collection Mini vMac on an iPad before finding that it didn't allow the use of internet resources and handled the keyboard incorrectly. He then compiled Basilisk II in Xcode on his MacBook (“a version of the macemu/iOS branch by user Jesús A. Álvarez”) and found that it worked on his iPad mini.

When he bought a Macintosh Classic II case on eBay, though, he discovered that the iPad mini's screen was too small, while standard iPads were too large and wouldn't fit in the case. “Apple also requires self-made software to be signed once a week which makes it very difficult to use.” The MacintoshPi project duly began.

Some six years on and it's finally available. What's more, it's already attracting a lot of attention. MacintoshPi builds on top of the Basilisk II emulator, allowing Mac OS 7 and 8 to run. It also uses the open-source SheepShaver emulator that dates back to 1998 to run Mac OS 9, making for a rather versatile retro Macintosh package.

To install it, users need either a Raspberry Pi 2, Raspberry Pi 3 or Raspberry Pi Zero 2 W (it won't work on the original model and it's not compatible with Raspberry Pi 4 Model B at the moment). “Right from the beginning, the goal was Raspberry Pi 3B+,” Jaromaz says. That's because he was able to draw inspiration from another project called BMC64.

“Raspberry Pi 3B+ is the best board for BMC64 – the fastest, low-level, bare-metal Commodore 64 emulator for Raspberry Pi,” he explains. Created by Randy Rossi and based on another Commodore 64 emulator called VICE, BMC64 offers true 50Hz/60Hz smooth scrolling and low latency between input and video/audio. “Randy optimised it for Raspberry Pi and omitted Linux entirely, hence the bare-metal,” says Jaromaz, adding that he wanted to follow the same path.

“It contains minimal and imperceptible latency and joysticks can be connected directly to the Raspberry Pi's GPIO, eliminating the additional latency of the USB interface.” It became a model for how Jaromaz wanted MacintoshPi to work. For good measure, he ended up including a dual-boot option that supports the Commodore emulator. It can mimic a Commodore 128 and Commodore PET too.

Hard work

To get MacintoshPi up and running, you simply need to compile and run a single script atop the legacy version of Raspberry Pi OS Lite. If you use the full legacy version of Raspberry Pi OS, you'd need to switch to the command line or set system start to CLI, but the process is neatly explained on Jaromaz's GitHub.

Getting the packages compiled and installed takes about two hours but the results are stunning. MacintoshPi bypasses the X.org manager and background tasks are reduced to a bare minimum (“the full resources of Raspberry Pi can be used,” Jaromaz enthuses).

MacintoshPi therefore boots and runs very quickly – if you put a Raspberry Pi running MacintoshPi next to a vintage Mac running the operating system, you'd be hard pressed to notice any difference. But the task has not been an easy one, as you can imagine, especially given the scope of the developer's ambition.

“After compiling emulators under the Raspberry Pi, I encountered a number of problems,” he says. “I was able to solve some of them after my experiences with these emulators acquired on MacBook and on iPads, but some required more work – searching for answers on forums or on non-existent sites using the Internet Archive.”

The work was complex. “I had problems with building the correct images, with configuring system images for internet support, with appropriate configs for emulators, the correct options for compiling emulators and, separately, SDL2, because SDL2 from a package won't work in emulation,” he continues.

“There were problems with performance, with the correct compilation of NetDriver, with versions of libraries (because they have to be legacy), with mouse movement delay and with sound.” You sense he could go on, but he has managed to go into battle and beat each and every issue. Indeed, you could replace a broken vintage Macintosh motherboard with MacintoshPi (connected to a replacement display) and carry on as before.

And that's what Jaromaz has done. After restoring the yellowing case of his Macintosh Classic II using a solution that's 18% hydrogen peroxide, and appropriate light exposure, he fitted a 10.1in Waveshare 11870 1,024 x 600 LCD display designed to support Raspberry Pi. With the screen turned upside down so that the wires were out of the way, he added a Logitech MK295 Silent Wireless keyboard/mouse combo, which proved attractive because they use a single Bluetooth dongle. He also added a couple of speakers to the Raspberry Pi's analog audio/jack input and a splitter, then popped them inside the case.

Up and running

But what happens at startup? “A very important part of the project is the MacintoshPi Launcher, which overwrites configuration files for the emulator and Raspberry Pi itself, reboots the Raspberry Pi already at a resolution equal to that of the emulated system, and launches the actual emulator with Mac OS, as soon as the Raspberry Pi OS boots,” Jaromaz explains.

“When starting Mac OS, the launcher – based on the resolution and selected system version – matches the appropriate startup chime (guessing, as it were, which Macintosh version the sound should come from) and plays that sound.”

When Mac OS is launched at the same resolution as a Linux system, there is no need for resolution translation and no associated mouse lag. “For example, in the emulation of Macintosh Classic at its standard 512x342 resolution, mouse lag does not occur at all, which would unfortunately be the case if the resolution on the Raspberry Pi OS was set to, for example, 1,920 x 1080,” he added.

“When you turn off the Raspberry Pi and restart it, it will still automatically boot the emulated Apple system. Only by exiting Mac OS and running a command such as **mac pios** or **mac cmd** would the Raspberry Pi OS (graphical or CLI version) be restored to automatic startup upon power-up.”

While that means it's possible to flick between operating systems, the main attraction is the vintage Mac OSes. MacintoshPi is perfectly capable of running games such as Lemmings, Age of Empires, Chessmaster 6000, Panzer General, Tetris and more across the three Mac operating systems. It will also fire up Netscape Navigator and Outlook Express and allow users to enjoy Telnet, Internet Relay Chat and Bulletin Board Systems via the Virtual Modem. Steve Jobs would be proud!

“Modem emulation is an interesting feature of this project,” says Jaromaz. “I applied the tcpser software used previously to Telnet BBSes within the VICE emulator, but the method of use seemed too complicated. Since I wanted modem emulation to work on emulated Mac OS systems as well, I decided to redirect the **tcpser** input to virtual device **/dev/tnt0**. Then the data can be read from another virtual device **/dev/tnt1**, which becomes a virtual serial port in the system.”

It's also possible to connect a real CD-ROM drive – a matter of specifying a path to the drive in the configuration file. CDEmu software allows any CD/DVD image to be plugged in and they'll be handled automatically.

“After connecting the image with **cdload image.toast** command, a virtual CD drive will appear on the desktop of the emulated Mac OS,”Jaromaz says. “Basilisk II and SheepShaver let you connect a floppy drive too, but I'm not sure how it would be mounted on Raspberry Pi.

“A workaround is to use the **dd** command and create an **disk.img** image of the floppy drive. Such an image can be plugged directly into MacintoshPi with a command such as **mac os7 disk.img**. After booting, a virtual floppy drive with the contents of **disk.img** image will appear on the desktop of the emulated Mac OS.”

Tapping into nostalgia

Of course, going to such lengths isn't for everyone. Some people merely want a quick burst of nostalgia or to get old software to run in the fastest possible time. If you're among this group, you can go to https://system7.app or https://macos8.app in a browser to enjoy another, recent Mac-related project called Infinite Mac which is also based on Basilisk II. As its creator, Mihai Parparita says: “It's a golden age of emulation.” And, if you're keen to relive the days of retro computing, the scene has never been so well supported.

“Nostalgia does play a big part in this – the opportunity to go back to the days of our youth,” says Jaromaz. “It's a chance to recall the software we used in our childhood and bring back those memories. But it also taps into curiosity. It's much better to read and understand biographies of famous figures associated with Apple Computer Inc, for example, knowing how the systems worked and looked at the time. Oh, and it's really great fun.”

And that, we have to say, is more than enough for us.

IT Pro created this content as part of a paid partnership with AMD. The contents of this article are entirely independent and solely reflect the editorial opinion of IT Pro

Once upon a time, supercomputers were rather different to the everyday systems used in business. They are still very specialised machines, but more in scale now than core technology. Even Frontier, the fastest supercomputer in the world and first Exascale system, is based on hardware that is not a million miles away from what you might find in a high-end gaming PC.

That means all levels of computing can learn something from these pioneering machines, but small- and medium-sized enterprises (SMEs) have specific lessons to gain.

Reaping the benefits

One factor that has affected all sizes of business over the last couple of years has been the switch to greater remote and hybrid working. An enabling factor here has been the growth in availability and sophistication of public, private and hybrid cloud infrastructure. This can provide the continuity backbone of software services that enables employees to enjoy the same employment environment experience whether they are working from home or in the office.

The underlying infrastructure for cloud computing can enjoy some of the same benefits as supercomputers through choosing appropriate hardware platforms. In particular, a denser server CPU platform can deliver a lower environmental impact and lower cost for the same level of service, or more compute power for the same money. If your remote workloads require GPU acceleration, this is now an option from some cloud providers too, and there are further benefits here from choosing an appropriate platform.

The latest supercomputers with Exascale capabilities or close to this have taken advantage of the greater density afforded by 3rd Gen AMD EPYC processors, which offer up to 64 cores per socket (so up to 128 cores and 256 threads per server) and up to 4TB of RAM. While this has benefitted the raw compute required for supercomputing workloads, it can also enhance the delivery of virtual machines and containers in a cloud environment. The more CPU cores a single rackmount server in a cloud infrastructure data centre can deliver, the more instances it can support, and it will also enable larger VMs with more cores.

With AMD’s platform price enabling servers with more high-performance cores for the money than competitors, and the lower energy running costs afforded by the higher density, cloud providers have been able to offer services with lower subscription fees than competitor platforms. Just as the latest supercomputers can deliver more compute performance with lower power consumption for the money by choosing AMD EPYC processors, cloud providers offer AMD EPYC CPU-powered instances with similar performance to other platforms but at a lower cost, making them a very effective choice.

For workloads requiring GPU acceleration, 3rd Gen AMD EPYC processors deliver 128 lanes of PCI Express 4.0, which enables the server hardware to support up to eight GPUs at full speed. The VMs can then offer workstation-grade graphics acceleration to remote virtual desktops. This has opened up the possibility of content creation in the cloud, enabling designers on the road to travel light and access a secure, powerful workstation environment using a relatively low-powered client system, even a notebook, over the internet.

Processor power

AMD has also just released its 4th Gen EPYC processor, which increases the maximum number of cores per socket to 96 from 64, further improving density and with 50% more memory, for a maximum of 6TB. This will be the first AMD EPYC platform to support DDR5, with twelve channels of 4,800MHz ECC RAM available. This will greatly enhance bandwidth, for even better performance when running memory-intensive workloads such as database services. The new processor will be available from AMD’s existing cloud partners, including AWS, Alibaba Cloud, Google Cloud, Microsoft Azure, Oracle Cloud, and Tencent Cloud.

Further down AMD’s roadmap, 128-core CPUs are due later in 2023. This will enable a dual-socket system to offer 256 cores (512 threads), twice as many as when using the current 3rd Gen AMD EPYC CPUs. For cloud providers, this will enable another step change in density, with the ability to offer high-performance instances at even more competitive prices compared to alternative platforms. Future supercomputers will similarly benefit from the greater density to deliver even greater compute power from a cluster.

Beyond everyday productivity, however, the ability to deploy AI/machine learning-empowered workloads is another area where emulating the latest trends in supercomputing could pay dividends for SMEs. There is a considerable emphasis on GPU power in supercomputers able to achieve Exascale performance levels, with Frontier for example getting most of its FLOPS from accelerators (AMD Instinct MI250, to be specific) rather than the host CPUs. Scientific workloads are being ported over to take advantage of this GPU power via environments such as Siemens CodeBench Lite, which facilitates support for the necessary software calls without the need for extremely low-level developing skills.

While GPU acceleration is taking these scientific workloads to a new level of performance, and enabling the use of much larger datasets, the focus on GPU power is also helping to accelerate the trend towards mainstream use of AI/ML. In the past, the need for expensive hardware acceleration has kept AI/ML within specialised domains that can benefit most greatly from its application. But the latest supercomputers offer an abundance of performance for this kind of workload, which is encouraging wider application. The greater affordability of GPUs and high-density platforms such as provided by AMD EPYC processor-powered servers have enabled cloud providers to offer instances tailored for AI/ML workloads. This is an area set to expand. With the pay-as-you-go model of the cloud, even SMEs will be able to consider AI/ML to empower analytics insights across datasets.

The computing revolution of the last few decades has been a consistent story of more performance for less money. This is the fundamental implication of the infamous Moore’s Law. But in the last couple of years, the fact that the fastest supercomputers are employing scaled-up and optimised variants of similar core platforms to more everyday systems has created more of a continuum of capabilities. Now, innovation in software and approaches at the very highest level can filter down more easily to SMEs. Particularly through cloud service providers, SMEs can gain benefit from powerful virtual machines running on supercomputer-grade hardware for affordable operating costs, as well as starting to take advantage of the burgeoning adoption of AI/ML.

IT Pro created this content as part of a paid partnership with AMD. The contents of this article are entirely independent and solely reflect the editorial opinion of IT Pro

IT Pro created this content as part of a paid partnership with AMD. The contents of this article are entirely independent and solely reflect the editorial opinion of IT Pro

Supercomputing entered a new era in 2022, with the inaugural Exascale cluster coming online. Oak Ridge National Laboratory’s Frontier was the first system to arrive that is able to deliver over 1 ExaFLOP of compute performance, or one quintillion floating-point operations per second. This will unleash a new world for research, including the application of AI/ML on a much larger scale. But it has significant implications for medium and large enterprises as well, particularly while they address digital transformation.

Frontier is an immense system, with its HPE Cray EX cluster composed of 9,408 3rd Gen AMD EPYC 7A53 processors sporting 64 cores apiece, for a total of 602,112 CPU cores. Each of these single-processor nodes is equipped with four AMD Instinct MI250X GPUs, making a total of 37,632 accelerators. While the CPU compute power is already enormous, the GPUs deliver the lion’s share of the FLOPS, so to take advantage of the compute power on offer from Frontier, workloads need to be enabled for GPU acceleration, including unique new features.

The AMD Instinct MI250X GPUs deployed in Frontier offer a range of capabilities, but two stand out. First is memory coherence, where the accelerator memory and system memory can be treated as a continuum, rather than requiring two copies of the data as in other standard systems. While it was already possible for multiple GPUs in a node to share their memory, facilitating the handling of very large datasets, adding system memory as well further expands the possibilities, particularly when each node can support up to 4TB of RAM.

What results is memory space savings that allow the processing of larger and more complex datasets. The significant reduction in programming code makes the code more nimble and efficient, cutting down on the time it takes to execute commands.

The AMD Instinct MI250X is also the first GPU to offer built-in networking, which enables distributed processing across nodes. When GPUs are plugged directly into the interconnect network, the communication between compute nodes or GPUs can become faster and more efficient, with smaller latency overhead. This will enable even more abilities to operate on huge datasets, which for some workloads will mean a qualitative change, not just a quantitative one. Certain kinds of insights that were restricted by scope will become possible. The scientific community researching cosmology or extremely complex environmental systems with Computational Fluid Dynamics will be the first to benefit from this.

While Frontier is the first supercomputer so far to take advantage of the AMD Instinct MI250X’s built-in networking, it’s not a special feature only for this system. This capability is likely to become available on a wider basis in more generally available HPC data centres, and GPU-enhanced code will take advantage of it to improve performance scaling. For the time being, the huge datasets used in certain domains within science will be the main beneficiaries of this technology. But all areas of data analytics can benefit from volume to deliver more effective insights. Any company needing to analyse the commercial behaviour of its customers on a global basis is going to have a huge amount of information to process. The ability to work on larger sets covering longer periods can deliver better actionable analysis.

Power consumption is another area where supercomputers like Frontier and Europe’s recently commissioned Lumi lead the way. Their choice of hardware delivers unprecedented levels of performance and consumes a lot of power overall, but in terms of how much compute per watt they provide, they are the most frugal systems yet. This is going to be an increasingly important attribute for medium and large businesses, not just because of concern for the environment but for the optimisation of running costs.

There are now over 7 million data centres around the world. Energy company Engie has estimated that these accounted for 4% of world energy consumption and 1% of greenhouse gas emissions. The hunger for compute power is showing no signs of diminishing, particularly as Internet of Things devices proliferate and AI/ML workloads are deployed in more and more areas. This compute must be delivered in the most environmentally friendly way to ensure that the increase in demand doesn’t come with a prohibitive environmental impact or cost.

The density of supercomputers such as Frontier and Lumi enables them to consume less power for their performance capabilities, thanks to efficient processor and accelerator design, which provides the most compute for the watt. However, HPC data centres don’t just consume energy directly to power their CPUs and GPUs, but also for the cooling required to keep these components at optimal operational temperature. This is another area where the technology used in the latest supercomputers offers plenty to emulate.

Traditionally, data centres have required sophisticated, power-hungry air conditioning to keep them at optimal temperatures. Not only does this draw a lot of electricity but can also eject a lot of heat into the surrounding environment. Using cooling that relies on water and natural airflow rather than actively refrigerated air conditioning can dramatically reduce the power consumption and environmental impact. Data centres in hot dry climates such as the US western and southwestern regions have also been deploying “swamp cooling”, which relies on evaporation to provide a cooling effect. Not only are these systems much cheaper to install than air conditioning (around half the price), but they also consume less than 40% of the electricity.

The efficiency afforded by density is set to increase, too. The processor side of supercomputing has just had a huge injection of extra power with the release of AMD’s 4th Gen EPYC processor. This increases the number of cores per socket by 50%. The top CPUs now offer 96 cores, enabling a dual-socket server to deliver 192 cores. Although this increases the thermal design power by nearly 30% over the equivalent 64-core 3rd Gen AMD EPYC processor, with 50% more cores the compute per watt will go down, making this an even more environmentally friendly HPC platform.

Medium and large enterprises have been undergoing a concerted period of digital transformation, which shows no sign of abating. The increasing use of AI/ML and data analytics in business practice entails a parallel escalation in the demand for HPC compute infrastructure. The supercomputing revolution, with its focus on GPU acceleration, density, and power consumption, shows the way forward. By borrowing ideas from the fastest computers in the world, medium and large enterprises can ensure they can address digital transformation in the most environmentally friendly and cost-effective way.

IT Pro created this content as part of a paid partnership with AMD. The contents of this article are entirely independent and solely reflect the editorial opinion of IT Pro

A major hardware organisation has implied that Nvidia is at fault for the widely reported instances of melting power cables on its new RTX 4090 graphics cards.

Nvidia's latest line of GPUs was released in October but their power cables have been mired in reports of extreme thermal faults leading to some users’ cables smoking or melting entirely.

Specifically, the GPUs' 16-pin 12VHPWR power cable, which adopts the new ATX 3.0 PSU standard released this year, has been identified as the core issue at play. Nvidia investigated the circa 50 extreme cases reported to it and concluded that “a common issue is that connectors are not fully plugged into the graphics card” - an issue of user error.

However, the Peripheral Component Interconnect Special Interest Group (PCI-SIG), a tech consortium which created the new power standard, and whose board includes representatives from companies including AMD, Arm, IBM, Intel, and Nvidia itself, has sought to distance itself from Nvidia’s claims.

It stated firms are responsible for rigorously testing their own hardware, and referenced an affected individual who has filed a class action lawsuit against Nvidia in a California district court.

"PCI-SIG wishes to impress upon all members that manufacture, market, or sell PCI-SIG technologies (including 12VHPWR connections) of the need to take all appropriate and prudent measures to ensure end-user safety, including testing for the reported problem cases involving consumers as alleged in the above-referenced lawsuit," the group said in a statement to the press.

“Members are reminded that PCI-SIG specifications provide necessary technical information for interoperability and do not attempt to address proper design, manufacturing methods, materials, safety testing, safety tolerances or workmanship. When implementing a PCI-SIG specification, members are responsible for the design, manufacturing, and testing, including safety testing, of their products.”

In the lawsuit (PDF), the plaintiff alleged that Nvidia “marketed and sold the RTX 4090 with a defective and dangerous power cable plug and socket, which has rendered consumers’ cards inoperable and poses a serious electrical and fire hazard for each and every purchaser”.

PC hardware reviews channel Gamers Nexus replicated the fault through the improper connection of the connector to the graphics card, and recorded the card reaching temperatures as high as 250ºC.

The group analysed cables sent in by affected viewers and determined that wear and tear proved the cables had been inserted at an angle, or with pins loose.

Despite this, in lieu of an audible ‘click’ when the cable is inserted properly, the reviewers noted that it’s very possible for a user to accidentally install the card insecurely. They also warned that those tinkering with cables after hearing about the issue might be opening themselves up to failure.

The 12VHPWR is a new power standard and used in Nvidia’s RTX 4080 and 4090 cards in place of the two eight-pin connectors that were used with the firm’s older cards. Rival AMD still utilises this older solution, having opted against including the 12VHPWR on its newest cards.

PCI-SIG’s statement is seemingly aimed at quashing fears that the thermal problems are endemic to the new standard.

IT Pro has approached Nvidia for comment.

IT Pro created this content as part of a paid partnership with AMD. The contents of this article are entirely independent and solely reflect the editorial opinion of IT Pro

At the dawn of the information technology revolution, supercomputers were the only computers, and some experts even thought things would remain that way. Even after the personal computer arrived, supercomputers remained a separate breed for some time. Now, however, there is a much closer relationship between desktop computers and the underlying hardware of the fastest systems in the world. This means that there are lessons to be learned and benefits to be gained from developments across the board.

Of course, there is still a huge performance difference between the system on your desk or in your company data centre versus the fastest around, even if the smartphone in your pocket has more processing power than was used to land a man on the moon. Supercomputers are now entering the ‘exascale generation’, delivering exaFLOPs of processing power, meaning they can calculate over 1018 IEEE 754 Double Precision (64-bit) arithmetical operations per second. The cost of these systems is beyond the reach of most organisations, in the hundreds of millions of dollars. They tend to be the domain of huge universities, governments and multinational research institutions, with resources shared out to the most ground-breaking scientific projects on the planet.

But the basis of these systems in many cases is now not so far away from corporate servers, workstations and even home desktops and notebooks. The first exascale supercomputer is the recently commissioned Oak Ridge National Laboratory’s Frontier, which boasts 136,408 AMD EPYC 7A53 CPUs with 64 cores apiece. These are special ‘Trento’ variants of the 3rd Gen AMD EPYC processor sporting the Zen 3 core, so part of the same ‘Milan’ generation that has been rolling out in data centres and cloud services recently. GPU acceleration for Frontier also comes from AMD, in the shape of the Instinct range, which isn’t a million miles away from the GPUs employed by end-user applications either – only specialised for GPU compute.

In Europe, the fastest supercomputer in the region (and third on the global Top 500 list) is LUMI, but its story is even more interesting because not only is it powerful, it has also been designed from the ground up to have the least possible environmental impact. It can deliver up to 550 petaFLOPS from its 200,000 cores, but it is powered by green hydroelectric energy. The heat it generates is captured and used to warm buildings in the area. This will be useful, considering that LUMI is located in Finland, well known for its icy winter weather. Considering COP26 agreements and the growing need to tackle climate change, as our supercomputing and general computing needs expand, it’s essential that we consider the impact this is having on the environment.

The LUMI supercomputer is Europe’s step towards its own exascale power, which is likely to be realised with JUPITER (Joint Undertaking Pioneer for Innovative and Transformative Exascale Research), which will be installed at the Supercomputing Centre at Jülich in Germany. The EU is setting aside €500 million for hardware, installation, and infrastructure. JUPITER is expected to come online after 2024, taking pride of place alongside a host of European supercomputing facilities.

The hardware inside JUPITER hasn’t been confirmed yet, but AMD’s processors could be in the running as they have been winning many of these projects due to their efficiency. Currently, five out of the top ten fastest supercomputers use AMD processors, and ten out of the top twenty. Putting this efficiency in perspective, before Frontier came along, the fastest supercomputer in the world was Fugaku in Japan. It has an average power consumption of 29MW. Frontier hasn’t been measured with the Top 500 list’s High-Performance Conjugate Gradient (HPCG) benchmark yet, but in the main Top 10 list it delivers over three times the Rpeak score and nearly three times the Rmax. Yet it only consumes 19MW, showing just how efficient its AMD EPYC-powered hardware is compared to Fugaku’s Fujitsu A64FX 48-core ARM processors. JUPITER aims to deliver exascale computing with just 15MW of power consumption.

Now that energy pricing is increasing fast, and power usage has become a major environmental concern, this kind of efficiency is paramount for the future of high-performance computing (HPC). Frontier delivers nearly five times as many petaFLOPS per MW as Fugaku, providing much cheaper, greener supercomputing. But HPC and even more everyday server applications can emulate this efficiency by adopting some of the approaches of supercomputing. The core density and efficiency of AMD EPYC processors are also available in regular pedestal and rackmount formats.

There are other benefits to supercomputers using x86-compatible CPUs like AMD’s EPYC, too. Four of the current top ten supercomputers employ non-x86 processors, but to take advantage of their tremendous processing power it’s necessary to customise applications extensively to run on their hardware. In contrast, supercomputers like Frontier and LUMI run x86 code, so while they have much greater scale and specific optimisation necessary to take advantage of this, this can be derived from software originally aimed at more modest commodity hardware.

Researchers can try out their workloads on departmental servers or local data centres to iron out the bugs before letting them loose on the supercomputer. Indeed, in the run up to the arrival of Frontier, this is what researchers have been doing, via the Center for Accelerated Application Readiness, which has provided access to test nodes for eight scientific teams. They were able to test their workloads including for astrophysics, molecular dynamics, photon science and biological systems.

This commonality of code also encompasses the supercomputer GPUs, which like the CPUs have enough in common with less specialised hardware to make it possible to test workloads on non-supercomputer hardware. Software such as AMD’s HIP development environment and ROCm drivers even make it possible to bring GPU code across from competitor hardware with modest amounts of effort.

Supercomputer code based on workloads such as computational fluid dynamics (CFD) or machine learning with TensorFlow can deliver incredible results, for example calculating vehicle aerodynamics and AI models respectively. These results can be replicated at a smaller scale on less potent systems using code from the same family. What is learned about these modern tasks in the supercomputing environment can pay huge dividends even for those without access. Supercomputers are also being deployed to harden cybersecurity, testing scenarios and finding protective measures that can be rolled out across governmental and commercial organisations.

At their birth, supercomputers were deployed to handle workloads that were not possible on any other device. They still have unprecedented scale compared to desktop systems and regular data centres. But there is much more of a continuum between the two realms than ever before, with particular emphasis on the environmental impact of computing. The efficiency of the latest supercomputers is available to every level of computing, thanks to similar architectures being available for smaller data centres as well. We can also benefit from software aimed at supercomputers that can be run on generic hardware – and vice versa. Rather than just being rare, specialised machines, the supercomputing revolution is providing learning across the full spectrum of data-processing activities.

IT Pro created this content as part of a paid partnership with AMD. The contents of this article are entirely independent and solely reflect the editorial opinion of IT Pro

The Japanese food giant Ajinomoto has recorded a record-breaking share price after receiving a boost from its technology used to make materials for the semiconductor industry.

The company, famous for its MSG seasoning, saw its share price reach a high of ¥4,357 ($31.46) on 29 November. This beats the company’s former record of ¥4,350 which was recorded back in March 1987, as reported by Nikkei Asia.

The company first expanded into the semiconductor industry in the 1990s, when it repurposed its amino acid technology to develop Ajinomoto Build-up Film (ABF), a microfilm insulation for electronic devices.

This was first adopted by a major semiconductor manufacturer in 1999, but its popularity has surged in recent years due to the chip crisis, especially since ABF has been in scarce supply since the pandemic began.

The company’s non-food segments are predicted to provide over 40% of its overall profit this financial year. Its net profit is on track for a 10% increase to ¥83 billion too, with the semiconductor material production area of the business helping to inflate its earnings the most.

Spurred on by this, the company revealed that it will accelerate the expansion of its ABF technology and may potentially invest more than the planned ¥17 billion ($122 million) to keep up with demand, said Taro Fujie, CEO at Ajinomoto, according to Bloomberg.

Ajinomoto forecasted that its ABF product shipments will grow 18% each year to March 2026, and the growth is predicted to stay in double digits until 2030, said Fujie.

Around 70% of the multinational food company’s ABF goes into data centre servers, and it’s expected that the expansion of 5G will also ramp up demand. It also provides electrical insulation for high-performance CPUs.

The company said that sales in the ABF unit grew 30%, reaching ¥37.2 billion, during the six months ending with September. Ajinomoto’s frozen food business, on the other hand, makes up around 20% of its revenues and reported a loss of ¥300 million in the same time period.

ABF can be found in most of the world’s personal computers, Ajinomoto said. The material is used by most modern chipmakers when designing the smaller components inside CPUs and GPUs.

Ajinomoto said that the need for CPU substrates grew in the 1990s for three reasons. This is because there was a rise in the integrations of CPUs in personal computers, PCs transitioned from MS-DOS to Windows operating systems, and there was an increase in terminals from around 40 to over a thousand today.

"Advances in circuit integration have made possible CPUs composed of nanometre-scale electronic circuits," the company said. "These circuits must be connected to the millimetre-scale electronic components in electronic equipment and systems. This can be accomplished by employing a CPU 'bed' composed of multiple layers of microcircuits, known as a 'build-up substrate'."

It said that the ABF helps to create the micrometre-scale circuits since its surface works well with laser processing and direct copper plating. It's now considered to be an essential material for creating circuitry which guides electrons from nanoscale CPU terminals to the millimetre-scale terminals on printed substrates.

If you buy a new computer today, it may well be using DDR5 memory. As the name suggests, this represents the fifth generation of the double data rate (DDR) RAM standard, originally laid down back in the year 2000.

Like DDR5 RAM, DDR4 is a versatile standard that officially comes in seven flavours, dubbed DDR4-1600, 1866, 2133, 2400, 2666, 2933 and 3200. You might assume that the numbers indicate the speed at which the different variants run, and that’s basically right – although since DDR transfers data at twice the speed of the internal clock, a DDR4-1600 module actually runs at 800MHz, and so forth.

Take a look at DDR5 options, and you’ll see that they start at DDR5-4800 and go all the way up to DDR5-7200. That’s an enormous increase, with the top frequencies more than doubling – but that doesn’t necessarily mean that business laptops fitted with DDR5 RAM will run twice as fast as those fitted with DDR4 RAM units.

How does DDR5 RAM work?

The chips in a RAM module are fast, but they’re not capable of reading or writing a full 64-bit chunk of data every clock cycle. If you look at the specifications of modern DIMM dual in-line memory modules (DIMMs), you’ll see that each has a specified set of timings that indicate how many clock cycles it actually takes to store and fetch data. These are normally written as a series of four numbers, such as 16-20-20-34.

To make sense of these numbers you need to know that RAM is arranged as a virtual grid: the latter three numbers tell you how many clock cycles it takes to select a specific “row” of memory for reading, and the first one tells you how many cycles it will then take to retrieve data from the desired “column”.

This initial value is called column address strobe (CAS) latency and, along with the clock frequency, it’s normally the most significant factor in determining the relative speed of a RAM module. That’s because most RAM operations involve reading a series of values from consecutive columns within a row, so the overhead of row selection comes up relatively infrequently – whereas CAS latency applies to every value that’s accessed.

Thus, a DDR4-3200 module with the timings above would be able to provide one chunk of data every 16 clock cycles; since the clock runs at 1.6GHz, that means it takes 10ns for the RAM module to complete a request for data from a specified location in an active row.

You can easily check on the timings of your own RAM, either by checking the BIOS or by installing a free tool for Windows such as HWiNFO64. This offers a table showing the timings and clock speed of each DIMM in your system, while a summary reveals the total installed memory, type, clock speed and channel mode – which we’ll discuss below.

Ideally, for the best performance, you want a high clock speed and low timing figures. Unfortunately, while it’s comparatively simple to make memory modules run at ever-higher frequencies, it’s much harder and more expensive to make the actual memory cells respond more quickly – so faster DIMMs tend to have higher latencies, at least in terms of clock cycles. As a result, the difference in performance between DDR4 and DDR5 modules is far smaller than you might expect.

How much faster is DDR5 RAM vs DDR4 RAM?

It’s impossible to simply say that DDR5 is "x% faster" than DDR4 modules. Even two DDR4 modules can perform very differently. However, with a spread of tests, we can get a sense of the range of performance offered by the two RAM families.

Sister site Tom’s Hardware recently did just this, carrying out a series of benchmarks on high-end modules running at different speeds. They found DDR5 had a big technical advantage, with faster frequencies directly translating into increased bandwidth.

Before you rush to ditch your DDR4 system, however, remember that everyday computing tasks may spend a fraction of their time waiting around for memory access. Indeed, since the timings for DDR5 are much higher than for DDR4, workflows that involve continually switching between RAM locations may see little benefit from the newer memory technology.

To measure real-world performance, Tom’s Hardware carried out separate tests using real-world applications such as Adobe Lightroom, Adobe Photoshop, Cinebench R23, HandBrake and Microsoft Office. It found the actual performance benefit of DDR5 is nowhere near as pronounced as synthetic bandwidth comparisons would suggest. While the fastest DDR5 supposedly offers nearly three times the bandwidth as the slowest DDR4 tested, the real-world margin was closer to 25%.

That’s certainly not to be sniffed at – for heavyweight data-processing tasks, a boost of that size is well worth having. However, a single 16GB DDR5-6400 DIMM will currently set you back around £250, while a DDR4-21333 module in the same capacity can be had for £60 – so whether the upgrade is worth the money is another question entirely.

Power consumption in DDR5 RAM vs DDR4 RAM

Clock speeds aren’t the only changes between generations of DDR. As manufacturing processes have improved, DDR data densities have increased, enabling computers to support ever-growing RAM allocations without requiring huge numbers of memory slots, or dozens of chips soldered onto the motherboard.

With DDR4 memory, a single RAM chip can typically store up to 2GB of information; since DIMMs normally use a maximum of eight chips per module, that means you won’t normally see a DDR4 DIMM larger than 16GB. DDR5 lifts the bar right up to 8GB per chip, paving the way to storing 64GB on a single DIMM.

Then there’s power management. Older DDR3 DIMMs required a standard power supply of 1.5V, but DDR4 reduced this to 1.2V and DDR5 drops it further to 1.1V. In practice, these changes don’t translate to significant reductions in power consumption, compared to the CPU and GPU. The actual amount of power consumed will anyway depend on how many chips are installed on the DIMM and how hard they’re working.

However, just as with CPUs, lower voltages enable memory chips to run faster without overheating. This means manufacturers can set higher operating frequencies – and users can push things even further to squeeze out a little extra performance.

How to boost DDR5 RAM performance

Testing your RAM

If you want to tweak the speeds and timings of your RAM, be prepared for a lot of trial and error, as it may not be immediately obvious if you’ve pushed too far. The odd single-bit glitch here and there might not have any noticeable effect until it happens in the middle of a system process and causes Windows to abruptly blue-screen.

If you’d rather not find out the hard way, you can use Windows’ built-in Memory Diagnostic tool to thoroughly test that your RAM is working as it should – just open the Start menu and type its name to find it.

The catch is that, because this tool needs access to the full expanse of installed RAM, it has to run outside of Windows – you’ll need to reboot the system and wait for the test to complete before you can get back to the desktop. Depending on how much memory you have installed (and how fast it is) this can be slow; a 30-minute wait isn’t uncommon.

Using XMP to boost DDR5 RAM performance

Every RAM module has its official frequency and timing specifications encoded into the hardware, using a system called Serial Presence Detect (SPD), but that doesn’t necessarily mean it’s limited to those speeds. The BIOS settings will normally give you a free hand to increase the speed and dial down the latency as much as you dare. The catch is that if you push too far beyond the official specification, you’ll start to get RAM errors, and sooner or later your whole system will most likely crash.

The XMP standard lets you overclock in relative safety; it allows memory manufacturers to embed additional profiles into their modules which go beyond the JEDEC-approved SPD values, to unlock the maximum speed that they’re capable of. For example, a DDR5-5600 module would normally run at 2.8GHz, but the manufacturer might decide, based on its testing, to offer an XMP profile that dials up the frequency to 2.95GHz. You can select this with confidence that the chips can run stably at these speeds. A second, more ambitious, profile might also be included, which could require extra cooling to run reliably.

Intel originally devised XMP, but it’s now supported by a wide range of motherboards – even many AMD ones. With DDR5 it gains an upgrade: while the XMP 2 standard used by DDR4 modules allows two profiles to be embedded, XMP 3 supports a third profile for additional flexibility, plus two user-configurable custom profiles, so you can experiment with your own system and store the memory settings that work best for you.

This might sound overly complicated, but the latest Intel chips (Alder Lake and newer) include a new feature called Dynamic Memory Boost, which switches automatically between XMP profiles according to workload. RAM is overclocked when the system needs a burst of data, then allowed to slow and cool off when things quieten down.

Dual-channelling with DDR5 RAM

One final factor that can affect RAM performance is whether it’s running in a dual-channel configuration. That may at first sound a little confusing, because all DIMMs are dual-channel by design – in other words, they behave like two separate memory modules working in parallel, offering double the bandwidth of the older single-channel SIMMs commonplace in the days of Windows 95.

On a modern motherboard, however, you can also install matching DIMMs in paired slots and run those in parallel, getting another two-fold increase in bandwidth. This is one reason why memory upgrade kits are often sold as pairs of DIMMs rather than as a single module.

Both DDR4 and DDR5 can be provisioned in a dual-channel configuration, but DDR5 takes the idea even further. Where DDR4 DIMMs use a single 64-bit data bus, DDR5 splits this into two 32-bit buses working together – so with matching modules you can take advantage of not two but three degrees of parallelism.

As we’ve mentioned, you can easily check whether your system is taking advantage of dual-channel RAM using a tool such as HWiNFO64. If it turns out that you’re in single-channel mode, that normally just means that your DIMMs aren’t identical. If so, it’s not a huge deal: dual-channel memory normally provides a real-world speed boost of less than 5%. If you’re already using unmatched DIMMs you’ll probably get a bigger benefit from adding another, rather than replacing existing modules to enable dual-channel operation.

As the renowned author, Douglas Adams, once wrote: “Space is big. Really big. You just won't believe how vastly hugely mind-bogglingly big it is.” It’s no real surprise, then, that getting to the bottom of the origins of the universe takes a massive – almost unbelievable – amount of computing power.

Getting to the bottom of some of the universe's deepest mysteries has meant setting up a massive computer, with Durham University among a handful of UK institutions jointly working on the project. Admittedly, it isn’t nearly as big as the Deep Thought system that features in The Hitchhiker's Guide to the Galaxy, and the answers are far less vague than “42”.

Before even beginning to tackle these questions, researchers sought a device powerful enough to peer into the very beginnings of space and time. This is where the James Webb Space Telescope (JWST) came in. Successor to the great Hubble Telescope, the JWST conducts infrared spectroscopy, with its high resolution and sensitivity allowing us to examine objects deep in space with detail its predecessor never managed. The JWST was launched into space last December and is already producing phenomenal results. Clusters of breathtaking images published by NASA speak for themselves.

This is only part of the journey, though, and institutions such as Durham University are hoping to use the tiny clues revealed by such images to help decipher the origins of the universe. They should, over time, begin to give answers to such questions as what the universe is, what it’s made from, what dark matter might be, and so on. But first, data needs to be crunched on a massive scale while, at the same time, researchers run in-depth simulations over and over.

The origins of everything

Durham set up a Cosmology Machine (COSMA) last year to begin answering these questions. This isn’t the first COSMA – that came online in 2001 – with the latest version, COSMA8, becoming operational in October 2021. It’s part of the Distributed Research utilizing Advanced Computing (DiRAC) facility, which has five deployments at the Universities of Cambridge, Durham, Edinburgh, Leicester and UCL. The DiRAC-3 memory intensive (MI) service at Durham runs workloads that require massive amounts of memory.

COSMA’s role is to run a simulation of the big bang. By tuning different input parameters and different models, taking into consideration things like dark matter and dark energy, the team attempts to match a simulation with what’s observed by astronomers using the JWST. Using a heavy amount of statistical analysis, the team can fine-tune the input parameters of these models to get a better idea of how the universe might have been formed.

To do that, COSMA8 has a high amount of memory per core, with the current system boasting 1TB of random access memory (RAM) per computer to power the cosmology simulation. This simulation which starts with the big bang before tracking how this might propagate through spacetime, can take months to run. The additional memory capacity helps to shrink that time span.

The compute cluster is based on Dell EMC PowerEdge C6525s. With four servers arranged in a 2U formation, each chassis contains 512 processing cores, with up to 3,200 MT/s (megatransfers per second) memory speed to decrease latency and PCIe Gen4 to transfer data quicker. The first installation had 32 compute nodes to iron out any problems with coding as well as testing and benchmarking. Shortly after, this was expanded to 360 nodes with plans to increase this even further to 600 nodes next year.

COSMA8 also features AMD Epyc CPUs, with 64 cores per processor. Each node contains dual 280-watt AMD EPYC 7H12 processors per node. According to Alastair Basden, HPC technical manager at Durham University, having so many cores per node decreases the amount of internode communication, speeding things up along the way. Basden’s team worked with Dell to move half of COSMA7 to a switchless interconnect, meaning all the nodes are connected in a 6D torus rather than to a central hub.

Simulation theory

One of the reasons for building this style of interconnect was that some of the code running the simulations didn’t parallelise 100%.

In order to scale well with the new technology deployed at Durham, Basden says the team there developed code in Swift to dynamically allocate tasks to nodes with available processing power, using task-based parallelism. This reduces the imbalance often seen in HPC codes. The code automatically rebalances itself periodically to consider different parts of the universe that may be denser or sparser. These simulations can take several months to carry out and produce petabytes of data which the team then analyses.

It was also important for the system to have a very fast scratch storage partition as when the simulation is running it needs to dump a lot of data to disk very rapidly.

With the system in place, Basden says it’s running as expected. The real challenge has not been with the hardware but with the software. For Basden, this means getting the software environment “working in a way in which maximises utilisation of the hardware and makes it easier for the researchers to use”.

Another challenge was in being one of the first to use direct cooling using water to take heat away from the system. He says that “was a bit of a learning experience in how to manage these systems”. He adds Dell helped the team thorough this challenge, and it is now very reliable, so much so that Dell will be offering direct liquid cooling across its next generation server portfolio when it launches next year.

As for the future, there is talk of the possibility of using composable systems from a company called Liqid that enables infrastructures to virtually pool GPUs, CPUs, memory, and network switches, among other hardware components. While Basden may not use GPUs any time soon – as he says CPUs are better for the memory intensive simulations Durham is running – it could be of use in allocating extra memory to a server when running those simulations.

This novel technology has led to Basden sparking up discussions with code developers, lab scientists and researchers on how they would be able to redesign their coding to take advantage of a much larger bank of cheaper memory.

US banking company Citi has announced its intention to transition the majority of its 300,000-strong global workforce to Snapdragon-powered mobile computing devices, citing battery life and AI features as key differentiators.

Speaking onstage at Qualcomm’s annual Snapdragon Summit event in Hawaii, the company announced that it would be moving more than 70% of its users over to Windows on Snapdragon products such as the Lenovo ThinkPad X13s.

“We heard from our employees about how AI made video calls more engaging by isolating human voices from distracting noises, blurring the background, and really addressing the needs of typical hybrid work environments,” said Patrick Javier, head of end-user computing for Citi. “With Snapdragon's dedicated AI engine optimising these processes, employees rave about how this enhances productivity without sacrificing performance or battery life.”

“Snapdragon also allows offloading of many other critical workloads required by bank IT, such as security, encryption, and communication onto specialised engines that sit directly on the SoC, again freeing precious CPU resources for the user or to further extend battery life.”

Based on the company’s Snapdragon 8cx Gen 3 platform, the X13s is a 13in device with a 1.6kg weight and a thickness of 13.4mm, including 5G and Wi-Fi 6 support, and up to 32GB of RAM. Lenovo also claims the device offers up to 28 hours of battery life, and Javier highlighted this as a key consideration in Citi’s hardware refresh process.

“At Citi, we are committed to sustainable progress and to advancing solutions that address climate change and support our net zero strategy. With Snapdragon-powered PCs, we found that for office productivity and collaboration workloads typical of our employees, that battery can last a full day on one charge,” Javier said.

“Across our global employee base, that means the elimination of thousands of metric tonnes of carbon and accelerating our transition to the low carbon economy.”

Partnerships have been a strong focus of this year’s event for Qualcomm’s Windows on Snapdragon division; the Summit highlighted the deepening of its relationship with Microsoft through bringing additional AI enhancements such as noise cancellation and background blur for video calls to Windows 11, and the launch of the new Microsoft Surface Pro 9 5G powered by the co-developed Microsoft SQ 3 Snapdragon processor.

In addition, the company also announced an expanded partnership with Adobe. The software company will continue to create native versions of its Creative Cloud programs for Windows on Snapdragon, with the latest additions being Adobe Fresco and Adobe Acrobat, joining its Photoshop and Lightroom products.

Qualcomm has officially unveiled the next iteration of its Snapdragon mobile CPU platform, the Snapdragon 8 Gen 2, promising a range of improvements with a particular focus on AI technologies.

Launched at the company’s annual Snapdragon Summit event in Hawaii, the Snapdragon 8 Gen 2 is ‘purpose built’ for AI, with an improved ‘AI Engine’ powered by an updated Qualcomm Hexagon processor.

Qualcomm said the chip has seen major architectural changes, such as the addition of a Tensor accelerator chip which is double the size of the previous generation, resulting in more than 4x faster AI performance and a 60% performance-per-watt improvement in power efficiency.

“We designed the [Snapdragon] 8 Gen 2 with one key technology: artificial intelligence,” said the company’s senior vice-president of engineering, Chris Patrick. “Qualcomm is the AI inferencing leader at the edge."

"Today we've already powered billions of edge devices across the world. We built on that expertise to make our new flagship an engine of innovation, a true AI marvel. We've integrated AI across the entire platform to usher in the new era of AI accelerated experiences.”

The new platform also includes an updated Qualcomm Kryo CPU with four performance cores and three efficiency cores, delivering up to speeds of 2.8GHz and 2GHz respectively.

A 3.2GHz prime core based on Arm’s Cortex-X3 technology also features, as does a new Adreno GPU with 25% improvements for both efficiency and speed. The company said this configuration offers up to 30% faster overall performance and up to 40% better power efficiency for longer battery life.

Another new feature introduced with this platform is the ‘Cognitive ISP’ - a physical link between the Qualcomm Spectra ISP and the Hexagon AI processor which uses neural network technology to automatically perform semantic segmentation.

This means that the camera itself can contextually recognise elements within a photo such as faces, hair, clothes and skies - and perform real-time adjustments to each one individually.

In addition to advanced on-device image processing, the AI functions supported by the new platform include multi-language translation, AI autofocus, and the world’s first support for INT4 on mobile, which essentially allows AI models to be trained on four-bit integers, rather than 32-bit floating-point integers.

According to Ziad Asghar, VP product management for mobile at Qualcomm, what this means is that more neural models will be able to run concurrently on the device.

“What we are providing is what we call Qualcomm AI Studio, which is a set of tools that allow you to take a model and actually shrink it down to integer four. And by doing that, now, instead of having to process 32 bits, you've got to process four bits only. That means you've got to do a lot less work, which means you've got to spend a lot less power.

“We are the first in the industry to do this,” Asghar said. “So it brings a huge advantage; a lot more AI is going to be able to fit in the same amount of processing, which is very, very powerful for us.”

Condensing 32-bit models to INT4 reduces the power consumption by a factor of 64, he said, and the company’s introduction of micro-tile inferencing also minimises the number of reads and writes for improved performance and longevity.

The company is aiming to attract more AI developers to the platform by unlocking these features, and is releasing a new developer kit to make it easier for them to start creating compatible machine learning applications.

The Qualcomm Innovators Development Kit will initially be made available to universities and research institutions before being rolled out on a wider basis in the future.

Qualcomm has also used these capabilities to add functionality to its Sensing Hub which uses dual AI processors to provide real-time contextual awareness by combining input from a range of sensors across the device, including cameras, microphones, gyroscopes, and more.

This will enable OEMs and developers to introduce a range of additional features, with some theoretical examples including the ability to identify the sound of glass breaking during the night to alert the user, and the ability to detect the orientation of the user’s face while lying down and prevent unwanted screen rotation.

Snapdragon 8 Gen 2 supports dual-active dual-SIM 5G, with a dedicated 5G AI processor to deliver better speed and stability throughput in both mmWave and sub-6GHz connections.

It’s also the first commercial chipset to support Wi-Fi 7 connectivity via the integrated FastConnect 7800 chip, taking advantage of the increasing number of Wi-Fi 7 routers coming to market. Qualcomm is boasting up to 5.8Gb/sec speeds, and less than 2ms latency, with cloud gaming and 4K streaming touted as key use cases.

Backing this up is support for the HDR10+ Adaptive standard, which compensates for changing lighting conditions when viewing HDR content, and OLED ageing compensation to keep OLED displays such as those found on the Samsung Galaxy S22 looking their best after prolonged use.

The Snapdragon 8 Gen 2 is also the first mobile chipset to support Unreal Engine 5’s Metahumans framework for creating and using hyper-realistic characters in games, and offers real-time hardware accelerated ray tracing, as well as 4K video playback at 60Hz.

Finally, Qualcomm’s AI tools are also being used to increase device security, including making facial unlock more secure by capturing data from facial unlock attempts, and routing it securely through to the digital signal processor (DSP) to perform detailed analysis on facial muscle movements and skin colour changes to ensure authenticity.

The new platform will be used in devices from major smartphone manufacturers including Oppo, OnePlus, Sony, and more, with the first Snapdragon 8 Gen 2 smartphones expected to hit shelves by the end of this year.

IBM has unveiled its new 433-quantum bit (qubit) processor, while adding three new companies to its quantum network.

The company said the IBM Osprey, its new processor, has the largest qubit count of any IBM quantum computing processor, more than three times the number of qubits in its 127-qubit IBM Eagle processor released in 2021.

The processor is expected to run complex quantum computations beyond what normal computers are able to. To underline its power, IBM said that the number of classical bits that would be necessary to represent a state on the new processor far exceeds the total number of atoms in the known universe.

Additionally, IBM revealed its IBM Quantum System Two, its quantum system that’s designed to be modular and flexible by combining processors into a single system, is expected to be online by the end of 2023.

The company said this machine will be a building block of quantum-centric supercomputing. The company’s goal is to scale up its systems to over 4,000 qubits by 2025, which it says will go beyond the capabilities of existing physical electronics.

“The new 433 qubit ‘Osprey’ processor brings us a step closer to the point where quantum computers will be used to tackle previously unsolvable problems,” said Dr Darío Gil, senior vice president of IBM and director of research.

“We are continuously scaling up and advancing our quantum technology across hardware, software, and classical integration to meet the biggest challenges of our time, in conjunction with our partners and clients worldwide. This work will prove foundational for the coming era of quantum-centric supercomputing.”

IBM has also added new companies to its IBM Quantum Network, made up of over 200 Fortune 500 companies, universities, laboratories and startups that are working towards a quantum future.

This includes Vodafone, to explore quantum computing and quantum-safe cryptography, the French bank Crédit Mutuel Alliance Fédérale, to find use cases in financial services, and Swiss innovation campus uptownBasel, to boost skill development and promote leading quantum and high-performance computing (HPC) projects.

The companies will have access to the world’s largest fleet of over 20 quantum computers which can be accessed through the quantum cloud, said IBM.

“As we continue to increase the scale of quantum systems and make them simpler to use, we will continue to see adoption and growth of the quantum industry,” said Jay Gambetta, IBM fellow and VP of IBM Quantum.

“Our breakthroughs define the next wave in quantum, which we call quantum-centric supercomputing, where modularity, communication, and middleware will contribute to enhanced scaling computation capacity, and integration of quantum and classical workflows.”

A year ago, IBM unveiled the Eagle processor, the company’s first quantum chip that surpassed 100 qubits. IBM developed a new 3D packaging architecture to build the Eagle, and the company said it could be used to develop more advanced quantum processors in the future.

Armari's Magnetar CXI-AD1000 shows what the 12th gen Intel Core i9 can deliver when optimised. The range-topping 12900KS variant here offers a 3.4GHz base clock for the eight P-cores with a 5.5GHz Thermal Velocity Boost frequency, while the eight E-cores can jump from 2.5GHz to 4GHz in the Max Turbo mode.

Armari gives the chip two speed-oriented partners in both the Asus ROG Strix Z690-F Gaming motherboard and 64GB of DDR5 5,200MHz memory. These come as two 32GB DIMMs, leaving two slots free for expansion up to the motherboard's 128GB maximum.

Graphics acceleration comes in the form of the PNY Nvidia Quadro RTX A4500. This combines a considerable 7,168 CUDA cores with 20GB of GDDR6 memory offering 640GB/sec bandwidth. This will be a great GPU for visualisation with excellent ability for co-processing acceleration.

Only one storage device has been included, although the 2TB Western Digital SN850 offers enough capacity unless you perform activities with heavy media usage. It's one of the faster M.2 PCI-E Gen 4 units, with 6,959MB/sec sustained reading in CrystalDiskMark 8 and 4,919MB/sec writing. If you need more storage, the Asus motherboard offers three more M.2 slots, while the Fractal Design Meshify 2 chassis supports up to 14 drives, although only six mounts are included as standard.

In the PC Pro benchmarks, this system delivered one of the fastest overall scores we have seen of 684, with just 32-core and 64-core AMD Ryzen Threadrippers ahead of it. The Cinebench R23 result of 28,167 shows just how much multi-threaded power this CPU has, despite only half its cores supporting Hyper-Threading.

In SPECviewperf 2020, the Quadro RTX A4500 graphics pushed this workstation to 160 in 3dsmax-07 and 495 in maya-06, showing excellent 3D animation modelling power. For more industrial activities such as CAD, engineering and product design, the system is equally adept. It delivered 129 in catia-06, 190 in creo-03, 469 in snx-04 and 299 in solidworks-05. And a LuxMark 3.1 score of 13,641 illustrates how great the A4500 is for GPU-accelerated rendering workloads as well.

The Armari Magnetar CXI-AD1000 is an exceptionally competent £4,500 workstation. The only thing to add would be secondary media storage if your work demands it. Otherwise, this system will despatch content creation workloads with ease.

Several cross-party senators have written to the US' security chief questioning Apple’s decision to use components from a Chinese chipmaker in its next iPhone.

US senators Marco Rubio and Mark Warner, vice chairman and chairman of the US Senate Select Committee on Intelligence, and Chuck Schumer and John Cornyn, sent the letter on Wednesday to the director of National Intelligence, Avril Haines. The letter called on Haines to launch a public analysis and review of Yangtze Memory Technologies Co (YMTC) and the risks it presents to US national security.

Apple acknowledged earlier this month it was reportedly considering procuring NAND memory chips for future iPhones from YMTC, which the letter said is a state-owned company with extensive links to the Chinese Communist Party (CCP) and its armed wing, the People’s Liberation Army (PLA).

The senators said Apple’s decision would introduce significant privacy and security vulnerabilities to the global digital supply chain the tech giant helps shape, especially due to YMTC’s ties to the CCP and concerning PRC-backed entities. They added the decision would affirm and reward the PRC’s disruptive and unfair trade practices, which undermine US companies by creating significant advantages to Chinese firms at the expense of foreign competitors.

“In July 2022, we wrote to Commerce Secretary Gina Raimondo to warn of the threat YMTC poses to US national security and to request that it be added to the Bureau of Industry and Security’s Entity List,” wrote the senators. “We made these arguments based on the company’s central role in CCP efforts to supplant US technological leadership, including through unfair trade practices.”

They underlined YMTC appears to have strong ties to the PRC’s military-civil fusion programme, as shown through its investors and partnerships. Its parent company, Tsinghua Unigroup, allegedly supplies the PRC military.

The senators also highlighted that in April, there were reports that alleged YMTC may have breached the US foreign direct product rule for supplying smartphone and electronics to Huawei.

The letter added policymakers have, for years, conveyed to the US public the importance of a competitive semiconductor industry to US national and economic security. It said a partnership between Apple and YMTC would endanger this critical sector and risk nullifying efforts to support it, jeopardising the health of chipmakers in the US and allied countries and advancing Beijing’s goal of controlling the global semiconductor market.

“Buoyed by a major contract with a leading global equipment vendor such as Apple, YMTC’s success would threaten the 24,000 American jobs that support memory chip production,” said the senators. “More broadly, such a partnership would also threaten the opportunities this market provides for research at US universities and further development of memory chips for civilian and military uses.”

The senators have requested a response from Haines by 1 October, 2022 and asked the director to produce a comprehensive public report on the Chinese company which can be used to inform federal agencies and the public as to the nature and risks associated with YMTC.

IT Pro has contacted Apple for comment.

SK Hynix will invest 15 trillion won ($10.9 billion) to build a new chip fabrication plant in South Korea, hoping to be ready for a chip boom that’s expected to come in 2025.

The South Korean company will build the new plant, M15X, in Cheongju, Chungcheongbuk-do, to prepare for future growth, it revealed today. The company is preparing for the next upturn in the chip industry expected to come in 2025 and hopes to be ready to provide more memory chips when the industry enters a boom again.

The new plant is set to be built from October on 60,000㎡ of land in the Cheongju Technopolis industrial complex which it hopes will be completed by early 2025. The 15 trillion won will be spent over the next five years to build the fab and set up production facilities. It’s expected to be a two-story building.

The company revealed the plant will make memory chips but will decide closer to the time whether they will be DRAM or NAND flash chips depending on market conditions, as reported by Reuters.

"Looking back on the past 10 years, SK Hynix could grow into a global company as it boldly carried out investment during crisis," said vice chairman and co-chief executive Park Jung-ho. "As we look to prepare for the next 10 years now, I believe starting the M15X will be a first step to lay foundation for a solid future growth."

Since becoming a member of SK Group, its parent, in 2012, the company has planned investments for the next 10 years. This was through its Future Vision which was the company’s pledge to invest 46 trillion won from 2014 to build three more fabrication plants. This included M14 in Icheon, M15 in Chengju, and M16 in Icheon. The company said it would decide a construction plan for the M17 fab, after reviewing the overall business environment including changes in the semiconductor business cycle.

SK Hynix noted that demand for memory chips is rapidly falling in the wake of the global economic slowdown and instability of the supply chain. However, it underlined that experts foresee that the business will begin to recover steadily from 2024 and rebound in 2025 as the memory business cycle becomes less volatile than in recent years.

This automatically-generated transcript is taken from the IT Pro Podcast episode ‘Solving the semiconductor shortage’. We apologise for any errors.

Adam Shepherd

Hi, I'm Adam Shepherd.

Rory Bathgate

And I'm Rory Bathgate,

Adam

and you're listening to the IT Pro Podcast. This week, we're heading down to the chip shop to talk about semiconductor supply chains and the recent global chip shortage. Our modern way of life has become completely dependent on technology. But this presents a bit of an Achilles heel. Semiconductors and silicon chips are the fundamental foundation on which all current systems are based. So what happens when those chips disappear?

Rory

Since late 2020, the tech industry has been struggling with acute difficulties in the semiconductor supply chain, affecting the availability of microchips, circuitry and other components. This has had a knock on effect on a huge range of products, including everything from laptops to Land Rovers, as availability plummets, and prices skyrocket.

Adam

So what's behind the supply chain challenges? And how close are we to resolving them? Here to talk us through the current situation as well as how the industry can strengthen supply chains in the future, is Ondrej Burkacky, Senior Partner at McKinsey’s Semiconductor Practice. Ondrej, thanks for joining us!

Ondrej Burkacky

Thanks for having me.

Adam

So let's talk about the chip shortage. One of the most obvious contributing factors was the impact of COVID on global supply chains. But do you think that was the primary cause, or were there other contributing factors?

Ondrej

I think that indeed it was, it had a very large contribution to what we are experiencing now. But it was for sure, not the only cause that we had, we also had some geopolitical tensions, mainly between US and China, that led to some restriction in global trading. That also led to several players putting semiconductors on stock, even before the shortage happened. And that combined with some individual occurrences, for example, the famous winter storm in Texas that happened in 2021. And also the fire in in one of the Japanese factories, were not really the I would say deal makers or breakers, but a near combination contributed to the situation that we are in now.

Rory

And outside of OEMs, what impact would you say the chip shortage has had on businesses?

Ondrej

Look, I think that, on the one hand, for sure, it is a shortage on the OEM side, right. But also when you look at the bit down and upstream there, the predictability of chip delivery has significantly suffered, right? Meaning also the tier one and tier twos are in a very unpleasant situation that on the one hand, they have a clear demand coming from the OEMs. But on the other hand, they are, they don't have a clear commitment from the semi compliance on suppliers, there is not enough supply existing here. And with that the whole supply chain starts to be very shaky, and less predictable. And also this invites to order typically called bullwhip effects, right, people starting to over order because there is a shortage, we'll just order more to get when there is an allocation, a sufficient amount; with that the demand for that explodes and so on and so forth.

Adam

So what level of shortfall are we talking about here? What's the discrepancy between supply and demand? Are we talking kind of orders of magnitude?

Ondrej

Yeah, it's a very good question. I think that that the answer to it is It depends really on the semiconductor type, right? And I think the one mistake that people typically do is like they say it's all the same, which is not true right. For example, we have almost no shortage on what is called leading edge technology, right, the CPUs and server chips. We also have less shortage on memory devices, being it storage, being it around for for compute, but we have a significant -meaning 30-plus percent - shortage of generic compute MCUs, right, that are there, for example, to be do the power management, that are there to operate the motor control in a dishwasher, in the washing machine, but also they are the ones that go into many, many cars to do motor control, airbag control, and all these all these things. So these are the ones where I would say the shortage is the most, and as I said, we are talking 30 perhaps 40% shortage versus what is the demand that is out there in the market.

Adam

So you mentioned laptops and server chips. We know that Intel in particular has been struggling with this, this issue, they've faced some strong headwinds as a result of various kind of supply chain issues; if CPUs themselves aren't being as affected as other types of of processors and kind of compute elements, what are the main challenges that companies like Intel are having to grapple with?

Ondrej

Yeah, look, I think that for for the whole PC or server, as you mentioned, right, it just doesn't only take one chip, right? There is memory in there, there is power management ICs there, right? And a bit ironic way, right? What caused the biggest headache for the server and PC OEMs were not the highest value semiconductors, typically, right, there was some shortage on the GPU side, but it was really this power management ICs, some of the power conversion ICs, some of the connectivity ICs. And it is like when you don't have this piece, you just can't ship the whole unit. So at some point in time, when you are the supplier of a CPU or or sent or other components that are not short, you also get affected by the shortage because people don't want to put more of on stock right of your equipment. Because they basically say, Well, I can't build the system anyway. So why should I have so much inventory on you? And with that, it's kind of there is the line definer, which is the component that is the most short, defines how much can get delivered, how much can get shipped.

Rory

So, what role would you say just in time manufacturing models have played into the current shortage?

Ondrej

Okay, I think that they did play a significant role. And you can see when you compare between different verticals, right, for example, when you take the consumer electronics industry, they typically do not have just in time delivery, because they have a high seasonality, right, I mean, there is a singles day, there is there is Christmas, Black Friday, Chinese New Year, where you have peaks in sales, and therefore you need to take precautions, and you need to put more parts on stock. If you take that other bookend here, which is the automotive industry, which has kind of optimised for just in time delivery. That is brilliant, as long as you have enough supply because then you don't need to have, have a lot of inventory. But this is one of the reasons why the automotive industry got struck so hard by the chips shortage is because they had basically no inventory along the entire value chain that they could survive from. And you can even see it when you compare between the different automotive OEMs. So if you look, for example, the Japanese players, they did a hard running from the from the Fukushima accident to put much more inventory in general on stock, right? So they basically being a bit, one of the inventors of just in time delivery, they actually did go away from it a bit. And so they were a bit hit less hard by the chip shortage than others, at least at the beginning; at some point, also that that inventory is used up, right?

Adam

So how well is the semiconductor supply chain recovering, at present? You're on site at a semiconductor fabrication facility at the moment; what is the current state of play in the kind of wider supply chain?

Ondrej

Yeah, I think it's it is a very interesting time here because of the demand jumped. Also, the plans of many semiconductor players to extend the manufacturing sites, right, have have been pulled in or extended. On the other hand, there is the physics here in place, right. So it still takes you minimum three years to ramp up a new semiconductor facility. And that's already a pretty short timeframe. Right. In many geographies, it's more five years that it basically takes you to get all the approvals to construct it, to bring in the tools to basically ramp up to a full yeilded product and with that, the players have focused very much on so called full vol expansions, which is basically expanding the existing sites, but that is almost already now exhausted, right. So, I think that there is like, if you if you find still a whitespace in an existing semiconductor fab, you are very lucky right? So the utilisation level there is already very high, right. So you are now looking into building new sites. And this is going to take time. And also there is one more consideration to put here out and that is that when you have such timeframes, like three to five years to build out a product production facility, you need to be damn sure that then not now you have to demand in place. Right. And this is why many people are now asking for customer commitments in a more mid or longer term range because nobody wants to end up having put five or 10 million in capex right to just find out that you know, three five years from now, the demand has softened and you cannot utilise that asset. And with that you're basically starting to lose money.

Adam

Yeah, absolutely. So how acute then are the shortages still? Are there still OEMs and manufacturers who are really struggling to meet their needs? And if so, how long will it be before semiconductor fabrication kind of supply chains are able to meet those needs effectively?

Ondrej

So then if I just take a very non scientific KPI and look into my calendar, how many meetings I have on chip shortage with how many players, I would say we are still in the middle of it, right. And I think this is indeed the case, right? Also if you look at the lead times, right, for the for the things that are really short. So our prediction says that 22 is going to be the same as 21, almost, right, only in 2023, we see also the first sites that basically start going piece by piece online that were basically built on top of what has been seen as the beginning of a shortage situation, which in the end means that by then we might get more chips that can be immediately used, it doesn't mean necessarily that only in 2023, all these disruptions in the supply chain are gonna go away, because it will take some time that restocking happens, it will some time that basically also the mix effect goes into the right price. We talked about this before that you need all the components to build a system in a way. So this can take well into 24 or even 25 until this gets fully sorted out. But there is ways for players right to optimise their sourcing, there is ways for them to commit to more longer term supply agreements. So where I would classify it to... So up till now, this being a very general problem for many players, right. And I think going forward, while there still might be a certain shortage, there will be ways for players to excel here and be able to source the semiconductors they need versus others might still struggle.

Rory

We're seeing governments become involved in the semiconductor shortage, America has just passed the landmark CHIPS for America Act that's providing around $54 billion in subsidies and the EU have their own Chips Act making headway. What kind of role do you see these these kinds of subsidies playing in the recovery of the semiconductor production? And do you think that governments have a real essential role in recovering for this industry?

Ondrej

Okay, I think that, generally speaking, the subsidies help in for companies to take the decision on making a CapEx investment, right, because very simply speaking, the return on investment, with a typical subsidy level, you get these days, reduces itself perhaps from nine years to even six, five years, right, depending on the on the level of subsidies you get. And with that, basically, your your, let's say, barrier to kind of overcome this investment decision and take the risk of perhaps not being utilised or you know, hitting a different market. And with that, it's reduced. So for that it is indeed, very helpful. Just one word of caution there for sure is, just because you have a Chips Act in place doesn't mean that the day after there is no more shortage, right? As I said before, there is still the physics of three to five years to build a semiconductor plant. Now also, with I think the very good intention that is out there, for some of the regulations it still only needs to become a bit more concrete, right, for individual players to be now sure that they will get a certain subsidies level. So they can bake it into their business case, and they can actually start executing.

Adam

So are there any ways in which the industry itself can look to increase the resiliency of the global semiconductor supply chain to try and prevent incidents like this from occurring in the future?

Ondrej

I think that this is really on the one hand, and important topic, on the on the other hand, a very tough one, right? Because the industry itself has always been optimised to be a global industry. Right? This is mainly due to the very high R&D investments that are required, right, the very high CapeX investment that you need, for example, on the manufacturing side, and the large scale effects you have right when you build things at scale, right. And with that, like when you look at the different, there are many segments right within the semiconductor industry. It's typically only a few players that act within these segments and they are always global, right? So they always produce for a global market. Now, one way is you can say okay, well, we can start to have more suppliers, right, we can start to become more resilient on that side, that will, by the way mean that the market shares of the players some way will be reduced here and there which might allow them to invest less in R&D, and so on and so forth. Right. So there is this challenge on that end. If you want to become more diversified and also more resilient, when it comes to certain regionalization, one really needs to be aware that this is a very long journey to go. Right. And there are also the subsidies that we talk about these days will not be by far enough, right, because we are talking typically about front end manufacturing, but this is just one piece, right? It does not include the wafers, does not include the speciality chemicals and everything, right. That's not including the back end manufacturing, so all these various steps in the value chain. So therefore, I think that something like full regional resilience is not anything that you can actually achieve and anyway, perhaps don't want to achieve, because it will just cause enormous amount of time and money. But really being more, having more transparency, where actually your main critical points are and being able to address some of them, I think is the strategy to go forward. And many companies are really exploring in the value chain, where these neurotic points actually are and whether they can do something about it.

Adam

Yeah. Cause the fact that, you know, a fire at one kind of Japanese manufacturing plant can have a notable impact on the entire global supply chain - that does seem like a pretty significant sort of single point of failure, in effect. And it does seem like greater diversification of the kind of manufacturing base within the semiconductor industry could potentially help to kind of mitigate the risk of that.

Ondrej

Yeah, I think Adam if I, if I may disagree with you on one thing, I think the headlines and the media presence of that fire were much larger than the impact it had.

Adam

Okay, interesting.

Ondrej

So, I think that again, right, it was one small contributor, if at all, due to shortage situation we have experienced, right. And and the simple proof for that is like that, the impact of that particular fire was kind of mitigated three, four months after it happened. And we still are in a shortage situation. Right. So I think this is this is indeed something which is not completely new to the industry. But indeed, what is newer to the industry is the overall high utilisation we are facing, right. So this is kind of the the fundamental problem. And then, as always, right, then any drop, right, that happens here, and there just has has a much larger contribution because we are just operating on the edge.

Adam

Yeah.

Rory

So Ondrej, you've brought up this minimum of three years to set up a new sort of chain of, a new factory, a new facility. But what does the outlook for the supply chain look like over the next, say, 12 months or so?

Ondrej

Okay, I think that, that, in the end, there is some more capacity now coming online, right? I said, over proportionally to the to the previous years. But still, I would expect that the shortage is going to be a big topic for the upcoming 12 months. As now, there is a lot of discussion on long term commitments, long term supply agreement, but this is nothing that basically helps you now for the next, let's say 12 months, right? This is more securing your volumes in 2024, 2025, right? Because you have to see as well, what are the production cycles here, right? So a typical wafer that travels through the frontend fabs takes perhaps 100 days, right, to get produced, then you have the back end operations take another two three weeks; add it to the mix, right? So in a way, the chips that we will receive in six months, perhaps five months from now are the ones that are produced now. Right. So with that you only have this very gradual change, and therefore we don't expect the shortage to end shortly.

Rory

So we had news last month, that SMIC have claimed that they've broken through with the seven nanometer chip. What kind of differences does competition on this level make to the overall supply chain? And how has it changed the outlook for their competitors?

Ondrej

I think that it's it's always good to have additional capacity right? And and I think also competition is healthy in a way that it basically drives companies to do more innovation, to become more productive. Particularly on the on the leading edge note, as I said, this was not necessarily the the biggest area of shortage, right? It's more about providing OEMs or fabless companies alternatives from where they basically can source their leading edge chipsets, right, rather than being super significant contribution to today's shortage. However, going forward, for sure, with a certain node migration, meaning shrinking of the chip structures always to the more leading edge nodes or certain types of, of chips, for sure, it's good to have capacity, more capacity on the leading edge node size. But just to be very clear here, right? And we are not talking about all chips becoming seven, five or three nanometer, right? If you take the example of a car, 80% of the chips that are in a car are actually 50 nanometer plus.

Rory

That's quite a stark difference.

Adam

Yeah, that seems large.

Ondrej

Indeed, it is. And I think therefore, we also do forget sometimes that when we look also on these chips programmes that are out there, right, we should not forget about mature technology and feature nodes. Because in the end there again to coming back to my my initial example, right, you might have too high sophisticated chip in your phone, in your smart device. But if it can't power up because the perceivably less sophisticated power management chip is not there, then you are out of power.

Adam

Yeah, I think there's a tendency to kind of, particularly in the tech and IT industries, to focus on the really kind of cutting edge stuff, on the smallest process technologies and the, as you said, the kind of bleeding edge manufacturing challenges, but a lot of the industry is not that.

Ondrej

Yeah, very right Adam, and a lot of the industry's, by the way, never gonna be it as well, right. If you look at for example, all the discrete semiconductors, so the ones that do power management, right, so AC/DC conversion, DC/DC conversion, right. All the nice things that we need for example, for electric cars right? There the differentiation comes less from a shrink of the node size, it comes more from the processing technology, switching from from silicon for example, to gallium nitride, silicon carbide is there the innovation driver right. So, one should not forget that there is not only one vector, right, you need to follow through for innovation in semiconductors and with the world becoming more broadly using a broader set of semiconductor technologies, right, with industry 4.0, all the IoT devices, etc. All of a sudden, all of these technologies get a boost in growth of demand and with that, they will also need a certain growth in supply to match that.

Adam

Unfortunately, that's all we've got time for in this week's episode, so we'll have to close out there but our thanks to McKinsey's Ondrej Burkacky for joining us today.

Ondrej

Thank you.

Rory

You can find links to all of the topics we've spoken about today in the show notes and even more on our website at itpro.co.uk

Adam

You can also follow us on social media as well as subscribe to our daily newsletter.

Rory

Don't forget to subscribe to the IT Pro Podcast wherever you find podcasts. And if you're enjoying the show, leave us a rating and a review.

Adam

We'll be back next week with more insight from the world of IT but until then, goodbye.

The tech industry has been forced to weather more than its fair share of storms over the past few years, but one of the most troublesome to navigate has been the ongoing issues in the semiconductor supply chain. The situation has resulted in delays, shortages and missed revenue targets for companies across a range of industries, and has prompted government intervention across the globe.

There are a number of factors behind the disruptions to the chip supply chain, and although measures are being taken to mitigate their impact, the crunch is set to continue. We’re joined on this week’s podcast by Ondrej Burkacky, Senior Partner at McKinsey’s semiconductor practice, to explain some of the reasons behind the crisis, as well as what’s being done to resolve the situation.

Highlights

“I think the one mistake that people typically do is they say it's all the same, which is not true. For example, we have almost no shortage on leading edge technology; the CPUs and server chips… but we have a significant - meaning 30-plus percent - shortage of generic compute MCUs, that are there, for example, to do the power management, that are there to operate the motor control in a dishwasher, in the washing machine, but also they are the ones that go into many, many cars to do motor control, airbag control, and all these all these things.”

“Our prediction says that 2022 is going to be the same as 2021, almost; only in 2023 we see the first sites that start going online piece by piece, that were basically built on top of the beginning of a shortage situation, which in the end means that by then we might get more chips that can be immediately used. It doesn't mean necessarily that only in 2023, all these disruptions in the supply chain are gonna go away, because it will take some time that restocking happens… So this can take well into 2024 or even 2025 until this gets fully sorted out.”

“What caused the biggest headache for the server and PC OEMs were not the highest value semiconductors, typically; there was some shortage on the GPU side, but it was really this power management ICs, some of the power conversion ICs, some of the connectivity ICs. And it is like when you don't have this piece, you just can't ship the whole unit.”

Read the full transcript here.

Footnotes

• US and EU thrash out plans to avert chip production “subsidy race”
• US companies have less than a five-day supply of semiconductors
• Incoming disaster or overblown: How will the semiconductor chip shortage affect enterprise IT?
• US and Japan double down on joint semiconductor research
• Western Digital and Kioxia to get £562 million funding in Japan
• Intel to acquire Tower Semiconductor for $5.4 billion
• China to establish chipmaking platform to secure domestic chip supply chain
• Chipmakers should brace for 'oversupply' in 2023 with crisis set to clear
• Texas winter storm halts semiconductor production
• Short circuit: Will the chip shortage end this year?
• Intel to invest €33 billion in EU chip manufacturing
• Europe proposes €43 billion Chips Act to tackle semiconductor shortage
• Intel pauses Ohio chip site development, citing delays in US CHIPS act subsidies
• Why India wants to become a chipmaking powerhouse
• Intel strikes $30 billion private equity partnership to fund Arizona plant expansion
• Intel warns customers to expect price hike for certain chips
• Can the UK semiconductor industry help solve the global chip shortage?

Subscribe

• Subscribe to The IT Pro Podcast on Apple Podcasts
• Subscribe to The IT Pro Podcast on Google Podcasts
• Subscribe to The IT Pro Podcast on Spotify
• Subscribe to the IT Pro newsletter
• Subscribe to IT Pro 20/20

Chip giant Intel has announced a partnership with Brookfield Infrastructure Partners, which will see the pair jointly invest $30 billion into expanding the firm’s Arizona plant.

Expected to close by the end of the year, the deal is a first-of-its-kind Semiconductor Co-Investment Program (SCIP) and represents a new funding model in the capital-intensive semiconductor industry, Intel said.

Under the terms of the agreement Brookfield will fund 49% of the $30 billion cost of expanding the company’s Ocotillo campus in Chandler, Arizona.

With a 51% stake, Intel will retain majority ownership and operating control of the two proposed new chip factories, which will support long-term demand and provide capacity for Intel Foundry Services (IFS) customers.

Announcing the deal, Intel CFO David Zinsner said the arrangement builds on the momentum from the recent passage of the CHIPS Act in the U.S.

“Semiconductor manufacturing is among the most capital-intensive industries in the world, and Intel’s bold IDM 2.0 strategy demands a unique funding approach,” he said.

“Our agreement with Brookfield is a first for our industry, and we expect it will allow us to increase flexibility while maintaining capacity on our balance sheet to create a more distributed and resilient supply chain.”

SCIP forms a key component of Intel’s Smart Capital approach, a framework designed to adjust quickly to opportunities in the market while also managing capital spending. Essentially, the structure allows Intel to access strategically-aligned capital to increase its flexibility and scale its manufacturing build-outs.

It’s the latest step in the company’s efforts to become less reliant on overseas manufacturers such as Taiwan’s TSMC and South Korea’s Samsung, following an extended period of supply chain disruption.

“We expect that SCIP, combined with the other pillars of our Smart Capital approach, will allow us to significantly accelerate our transformation and help deliver the more globally balanced supply chain the world needs,” Zinsner added.

The chipmaker said its partnership with Brookfield will enhance its balance sheet and allow it to tap into a new pool of capital below its cost of equity, while also protecting its cash and debt capacity for future investments.

The move is expected to boost cash flow by $15 billion over the next several years, while also enabling Intel to replicate the co-investment model with other partners in the future.

“By combining Brookfield's access to large-scale capital with Intel’s industry leadership, we are furthering the advancement of leading semiconductor production capabilities,” said Sam Pollock, CEO of Brookfield Infrastructure.

“Leveraging our partnership experience in other industries, we are pleased to come together with Intel in this important investment that will form part of the long-term digital backbone of the global economy.”

There’s no company more notoriously controlling than Apple. Even if your device has been on your desk or in your pocket for years, if it requires repair, you almost certainly have to go back to Apple or one of its authorised service centres to get fixed up. At least, that was the case. At the end of April, Apple’s new self-service repair programme went live in the US, and gave a glimpse of a world where anything is possible – even fixing your own iPhone.

Now anyone can log on to the Apple website and find repair manuals for the two most recent generations of iPhone, which detail how to safely take them apart and replace everything from the camera to the screen, without a trip to the Genius Bar.

“Creating greater access to Apple genuine parts gives our customers even more choice if a repair is needed,” said Jeff Williams, Apple’s chief operating officer (COO), in a press release when the programme was announced at the end of last year.

On the surface, it does seem like a major step forward for the so-called right to repair. “We were happy to see Apple make some move towards expanding repair options, that's for sure positive,” says Ugo Vallauri, co-founder of the Restart Project. However, though he is receptive, Vallauri doesn’t believe the new programme goes far enough. “When you look at their manuals, you see that it basically requires a whole lot of tools that are proprietary to Apple, that they will be the sole provider for,” he continues. “This is not the right to repair.”

Apple is still in control

Despite the apparent openness, he still believes Apple is exercising close control. “The biggest problem is not necessarily with the manuals, per se, but that Apple is pushing ahead with requiring part pairing for the parts,” says Vallauri. Under the programme, any new components requested will need the buyer to supply a serial number, which will be linked to the specific iPhone being repaired, limiting what can be done with spare parts.

Then there’s the question of proprietary tools. To help tinkerers buy the widgets and tools they require, the self-service repair programme has its own online store. But buyers beware, they may not be for the faint-hearted.

“It is a substantial amount of equipment, it might be [only] appropriate for a repair store,” says Vallauri, and he might have a point. For example, removing an iPhone screen with Apple’s approved method requires the use of a £209 “Heated Display Removal Fixture”, effectively a piece of industrial equipment. To install a new screen, a separate £176 piece of hardware called a Display Press is required. “It's not exactly the kind of situation where most people would just order a set of tools like that and perform a repair,” says Vallauri.

Even if self-repair is an option, the economics of it won’t make sense for ordinary users. “Is this a step forward in terms of repair options? Definitely. Is this something that truly competes with you going in an Apple store and getting the screen replaced by them? It’s unclear,” says Vallauri.

In fact, Vallauri speculates that although the move appears to be Apple throwing off its restrictive shackles, the structure of the new programme represents a consolidation of the market. “If you sum the cost of renting the tools and buying the parts, even if you were to receive a sum back for sending the broken parts back to Apple, it will still not necessarily be competitive with going into an Apple store,” says Vallauri. “At that point, the question becomes, is this a move towards controlling more of the repair market by creating a restrictive vision for what right to repair should be?”

Apple might be on the defensive

One read on the move by Apple is that although the new programme is being positioned as an egalitarian move, it’s actually a defensive one. The theory goes that it’s designed to pre-empt regulators in both the US and Europe, who are taking an increasingly sceptical approach to monopolistic or controlling behaviour of big tech. This is symbolised by the passing of the Digital Markets Act (DMA). The UK, too, has outlined a new right to repair law.

“We're seeing Apple and other manufacturers attempt to 'concede' to an extent, by giving a little bit more,” says Vallauri. “But this still differs quite significantly from empowering everyone to make a decision as to how and where and by whom a device should be repaired.”

So far, the service is only available in the US and covers just the iPhone 12 and 13, but it’s expected to launch in Europe later this year, with more Apple devices supported. So what would make it good enough for Vallauri?

“The right to repair that we’re advocating is for everyone to be able to choose who should perform the repair, on products that are designed to be repaired, with tools that are common and not proprietary,” he says. While this is a step in the right direction, he believes that the Apple self-repair programme is still in need of some patching up.

Samsung announced on Friday that it will spend roughly $15 billion on building a new chip research complex in South Korea by 2028, as the tech giant seeks to ease restraints on semiconductor scaling.

Casting light on the complex’s deliverables, Samsung said the new chip facility will be built at the company's Giheung campus to foster research into new technologies, including novel wafer fabrication processes for memory and system semiconductors.

Upon completion, the South Korean facility, which will stretch over 109,000 square meters at Giheung, will further strengthen the tech giant’s foothold in the semiconductor industry.

The Giheung campus also hosted Samsung's 1992 development of the world's first 64MB DRAM, which led the company to scoop the top spot among chip manufacturers the following year.

"We are taking on a new challenge from the very location where we broke ground 40 years ago to build our first semiconductor plant," said Jay Y Lee, Samsung vice chairman.

"If we hadn't made bold R&D investments for next-generation products and in products that came after that, there would be no semiconductor business for Samsung today. We need to continue our tradition of investing preemptively and emphasizing technology," added Lee.

As the global chip shortage continues to stress the big tech, Samsung is continuing to benefit from inflation resulting from the deficit. Increasing semiconductor prices helped Samsung achieve its highest quarterly profit since 2018, according to an earnings guidance published by the company late last year.

Additionally, Samsung announced in July that it would build 11 new chip manufacturing facilities in Texas’ Austin and Taylor, with an investment of $191.2 billion.

Fibre optic 10GbE networks are becoming affordable for SMBs, and TP-Link’s TL-SX3016F switch intends to make them even more appealing. This slimline 1U chassis offers 16 SFP+ 10GbE ports for a shade over £400, to deliver an incredibly low cost per 10GbE port of only £26.

You’ll need to factor in the cost of laser transceivers and fibre optic cabling, but a big advantage over copper 10GbE installations is they support much longer distances. Furthermore, unlike blue-chips that may lock you in with their own coded transceivers, the TL-SX3016F has no restrictions and happily accepted our PlusOptic 10GBase-SR models, which cost £35 each and support connections of up to 300 metres.

The TL-SX3016F has the capacity to handle a high data throughput, as it’s endowed with a speedy 320Gbits/sec backplane. Combine this with a high 238 million packets per second (mpps) forwarding rate and you have a switch that easily has the measure of more expensive 10GbE SMB products like the Zyxel XS1930-12HP.

A feature you wouldn’t normally expect to see at this price is dual redundant power supplies. Two sockets are provided at the rear, with the first PSU acting as the primary source; if that goes down, the second PSU steps in immediately to take over.

You also get a choice of management modes. The switch offers its own web browser interface or it can be integrated into TP-Link’s Omada cloud portal. In standalone mode, the web console sensibly requests the default administrative password to be changed and then presents a dashboard view showing the status of each port.

Tabs are provided for quick access to the various functions, with the switch offering an abundance of Layer 2 (L2) features, including port, MAC and protocol-based VLANs, and QoS traffic prioritisation, plus static and LACP link aggregation groups. VoIP networks are covered as the switch identifies traffic from IP phones using their organisationally unique identifier (OUI) and automatically prioritises it by dynamically creating voice VLANs.

The switch also offers basic Layer 3 routing capabilities. Referred to as “Advanced L2+” or “Layer 3 Lite”, it supports static IPv4 and IPv6 routing, but not the dynamic routing found in more costly full Layer 3 switches.

There are some nice surprises for switch monitoring. Along with support for all versions of SNMP, the switch has an internal sFlow agent for sending packet sampling data to a collector device. We found this very useful as we pointed the agent at a host running the WhatsUp Gold 2022 network-monitoring software, which promptly started reporting back on switch port utilisation, traffic throughout, the top detected apps and all endpoint conversations.

To remotely manage the switch, you’ll need to deploy TP-Link’s cloud controller at each site. TP-Link provides free Windows and Linux software versions, but for testing we use its preconfigured Omada OC300 hardware controller which costs around £110; the smaller OC200 costs only £63.

The cloud portal lists all assigned controllers; selecting one takes you to its own web interface, which provides a central location for managing and monitoring all TP-Link switches, routers, gateways and wireless access points. New devices initially appeared in the controller’s web console as “pending”, and once we had “adopted” the switch, it disabled its local web interface and took all settings from the controller.

The portal provided plenty of information about switch operations but some features, such as L3 options and the sFlow agent, can only be configured when it’s in standalone mode. This wasn’t a problem as we set these up before adopting the switch into Omada and set the portal’s override feature so all local settings were retained.

The TL-SX3016F is a great choice for small businesses that want a high-performance fibre 10GbE core switch and larger organisations looking for a low-cost aggregation solution. It partners a big helping of SFP+ ports with a great range of features, and all at a price the competition will be hard-pushed to match.

TP-Link JetStream TL-SX3016F specifications

3D printers have a range of uses within a business, whether it’s for printing custom circuit boards, scalable prototypes, or precision-moulded parts for a larger machine. Innovative companies have

Whether you’re making the step up to getting your business’ first 3D printer, or simply looking to further invest in the technology, there are a number of tools and accessories that can make your 3D printing experience that much better, regardless of the particular 3D printing use case.

Filament storage container

As anyone who has had the misfortune of having to deal with a 3D printer jam knows, it’s really important that moisture and dirt is kept away from filament as much as possible. Wet filament loses its structural integrity, and will not stick together properly in the printing process.

To that end, there are a range of filament storage containers available, to suit any size and filament type. Although a tightly-shut box could suffice to keep dust out, in wetter climates such as the UK it might not prevent moisture from getting to the filament held within.

This is where containers with desiccant inside come in. A desiccant is any compound that absorbs large amounts of moisture from the air, and they are often packed with products to absorb unwanted moisture. The most common day-to-day example of a desiccant is silica gel, which is often found in packets inside shoe boxes to keep moisture from compromising the quality of the shoes.

Desiccants can work wonders to keep filament dry, but after time will need to themselves be dried out or replaced altogether. Alternatives to using desiccants within filament containers include vacuum-sealed containers, which can efficiently keep filament in pristine condition, or vacuum packets into which conventional containers can be sequestered.

Filament dryers

If you are unable to keep your filament dry, it can be salvaged by being precisely dried. This is where filament dryers come in, offering consistent temperature and humidity control and specifically built with filament in mind.

The more premium filament dryers even have screens that show users temperature readings, and allow for settings to be tweaked to suit exact drying needs.

Digital Calliper

A digital calliper can take the accuracy of your measurements to a new level, providing accurate measurements for every occasion on a handy digital display.

Businesses that invest in a digital calliper can forget having to awkwardly mark a ruler to transfer measurements from one print to the next, with the most accurate models offering readings accurate to three decimal places.

Good quality print beds

A house is only as stable as the foundation that it's built on, and a 3D print is no different. Investing in a good quality print bed can elevate the sturdiness and accuracy of your prints,

Good thermal resistance is a must, as print beds can be heated to 90-110°C during the printing process. Heated beds help to ensure that the first layer of the print remains stable, to prevent the entire structure from sticking to the nozzle and potentially tearing itself apart as the print continues.

Print remover

Even the best quality print materials cannot totally guarantee that prints won’t become fused to the print bed once complete, and in these circumstances it’s important to have a print remover available.

Also known as ‘hot bed shovels’ or ‘spatulas’, print removers are essentially a sharp, flat blade that lets a user scrape their print from the bed without risking damage to the finished product. If you don’t fancy ordering a custom tool online, any good hardware shop will have a scraper/spackle knife in stock, also known as a putty knife. Although these are intended for plastering jobs, they work just the same as a print remover and are even available in larger sizes to suit your needs.

Range of nozzles

Nozzle attachments free you from the confines of your default nozzle width, allowing you to fill larger builds quicker or alternatively achieve a level of detail impossible with larger filament. This could be especially useful for businesses 3D printing custom tools, or even printing custom circuit boards.

Conductive filament

On the previous point, if your business is looking to 3D print circuit boards, 3D printers can go a step further than just setting out a blank board with holes for the conductive element.

Conductive filament can be used to lay down a predefined board pattern, and is made of materials such as carbon fibre and graphene. It’s important to note that 3D printed filament can only take quite a low voltage, so proposed builds must always take into account the manufacturer-defined limits to avoid damage or safety hazards.

Companies offering conductive filament include Protopasta, NinjaTek and Voxelab.

Nozzle cleaning tools

Needles make for ideal nozzle-cleaners, as the sharp and thin tips can get into the nozzle to clear away any excess filament that might be causing problems or even blocking the nozzle entirely.

For best results, it's recommended that you heat the nozzle to melt the excess filament as much as possible, allowing it to be easily scraped away with the sharp tip of the needle.

Specialist cleaning kits are available, with a range of needle sizes and proper grips to prevent unnecessarily fiddly cleaning jobs or even minor injury.

Computing giant Nvidia has revealed that its Q2 revenue came in lower than expected, finishing at $6.7 billion compared to its outlook of $8.1 billion.

The figure represents a 19% sequential drop but a growth of 3% over the same period of last year.

In an announcement, the company said the numbers reflect a reduction in channel partner sales, likely due to “macroeconomic headwinds”. Nvidia also implemented pricing programmes with channel partners to reflect challenging market conditions, which it expects to continue into Q3.

Data centre revenue fell short of expectations during the quarter due to supply chain disruptions but did manage to hit $3.81 billion, up 1% sequentially and 61% from 2021.

The overall shortfall, however, is largely down to weaker-than-forecast gaming revenue, the company said, which clocked in at $2.04 billion - down 44% sequentially and a 33% drop from the same period of last year.

Nvidia added that the full results are also expected to show $1.32 billion of charges, which are largely related to inventory and related reserve, based on revised expectations for future demand.

“The significant charges incurred in the quarter reflect previous long-term purchase commitments we made during a time of severe component shortages and our current expectation of ongoing macroeconomic uncertainty,” explained Colette Kress, EVP and CFO of Nvidia.

“We believe our long-term gross margin profile is intact. We have slowed operating expense growth, balancing investments for long-term growth while managing near-term profitability. We plan to continue stock buybacks as we foresee strong cash generation and future growth.”

Commenting on the firm’s decline in gaming sell-through projections, Nvidia founder and CEO Jensen Huang said the firm adjusted its channel prices due to its longer-term expectations.

“As we expect the macroeconomic conditions affecting sell-through to continue, we took actions with our gaming partners to adjust channel prices and inventory,” he said.

“Nvidia has excellent products and position driving large and growing markets. As we navigate the challenges, we remain focused on the once-in-a-generation opportunity to reinvent computing for the era of AI.”

Zyxel takes multi-gigabit switching to the next performance level with the deceptively small XS1930-12HP, which is positively bulging with features. Not only are all 12 ports on this desktop-sized unit the 10-gigabit (10GbE) variety, but the first eight provide 802.3bt PoE++ services.

The switch is incredibly versatile. All ten copper ports support 100Mbits/sec, 1GbE, 2.5GbE, 5GbE or 10GbE connections and you can easily see their operating speed as each one has a five-colour status LED. The last two are standard 10GbE SFP+ fibre ports for high-speed uplinks over long distances.

A generous power budget of 375W means the switch can easily provide 30W on all eight ports with enough headroom to power quite a few high-performance Wi-Fi 6 access points (APs) that need up to 60W. The LED status panel comes into play again here as it shows which ports are providing power, while a vertical light bar alongside reveals the total draw.

At its foundation, the XS1930-12HP is a Layer 2 (L2) switch with basic Layer 3 routing capabilities. Sometimes referred to as “Layer 3 Lite” or “Advanced L2+”, it supports static IPv4 and IPv6 routing, although not the dynamic routing found in more expensive full Layer 3 switches.

Standard L2 features are plentiful and include port and protocol-based VLANs, QoS traffic prioritisation plus static and LACP link aggregation groups. VoIP networks are covered since the switch identifies traffic from IP phones using their organisationally unique identifier (OUI) and prioritises it automatically by dynamically creating voice VLANs.

Then there’s management. You can opt for standalone mode or use Zyxel’s Nebula Control Center (NCC) platform to move it into the cloud. The local web console fires up a quick-start wizard that insists you change the default admin password and SNMP community name and, although it doesn’t provide any switch graphics, tables offer plenty of detail on system utilisation plus port and PoE usage.

If you’re concerned about exceeding the switch’s power budget, you can set priorities on individual multi-gig ports. Assign one of three priorities to each one and, if the drain reaches the total power threshold, those with the lowest priorities will be switched off first.

Moving the switch to cloud management is a cinch. We used the Nebula iOS app on an iPad to scan its QR code and assign it to our site. A minute later the switch appeared online in our cloud portal, received a firmware update and had configuration access in its local web console disabled.

Selecting the switch from the portal’s monitoring page presents a full status readout and location map, while a multicoloured graphic below shows all active ports, their connection speeds and those delivering power. Graphs are provided for network traffic and power consumption, but if you want more than a rolling 24 hours of data, you’ll need to upgrade the free NCC basic service to a Plus licence, which extends this to seven days.

The portal provides a range of global settings so any switch added to the site automatically receives them. These include QoS profiles, port mirroring and voice VLANs. Security can be stiffened up by limiting all management access to specific host IP address ranges and only allowing the switch to obtain its management address from authorised DHCP servers.

Individual ports on the switch can be configured from NCC, with options for enabling or disabling PoE services and fixing the connection speed, and a Plus licence enabling bandwidth controls. Up to five global PoE schedules can be created and assigned to selected ports so you can decide when powered devices such as wireless APs are active.

Its smart combination of high-performance multi-gigabit ports and generous PoE++ services puts Zyxel’s XS1930-12HP in a league of its own. SMBs that want to deploy the latest Wi-Fi 6 APs and team them with fast links to storage and servers won’t find a better alternative.

Zyxel XS1930-12HP specifications

Intel is reportedly close to sealing a $5 billion deal with Italy to build an advanced semiconductor packaging and assembly plant in the country.

According to Reuters, two sources briefed on discussions revealed that Rome is willing to fund as much as 40% of the chip maker’s total investment in Italy, which they said is expected to rise over time from the initial $5 billion.

The government of outgoing Prime Minister Mario Draghi is currently working to secure an agreement by the end of August, the unnamed sources added, which would see the deal secured before the country’s snap general election on September 25.

The new factories would leverage the latest technology to “weave together full chips out of tiles,” the news outlet reports.

Several Italian regions reportedly registered their interest in Intel’s investment, with Apulia, Lombardy, Piedmont, Sicily, and Veneto early contenders.

According to the sources, Intel and the Italian government have now shortlisted sites in two of the regions, with those said to be the northern regions of Piedmont and Veneto.

Back in March, Intel announced a €33 billion investment to go towards semiconductor research and development (R&D) and manufacturing in the EU, to include a rumoured €17 billion mega plant in Magdeburg, Germany.

It was also revealed the chipmaker was planning a new R&D and design hub in Plateau de Saclay in France, as well as a €4.5 billion factory in Italy. The company had also previously considered the UK but abandoned the plans following Brexit.

“This broad initiative will boost Europe’s R&D innovation and bring leading-edge manufacturing to the region for the benefit of our customers and partners around the world,” said Intel CEO Pat Gelsinger at the time. “We are committed to playing an essential role in shaping Europe’s digital future for decades to come.”

Under its Chips Act, the European Commission announced earlier this year that would provide around €15 billion in additional public and private investments by 2030, on top of €30 billion of public investments that were already planned.

On Rome’s side, the Italian government has reportedly ringfenced €4.15 billion to entice chip manufacturers to the country and invest in new industrial technology.

The move forms part of Intel’s drive to invest $88 billion in building resources across Europe as it looks to reduce its reliance on Asian manufacturers and avoid future supply chain issues like those seen in recent times.

As well as its investment in Europe, the chip maker also began construction on two new US chip factories back in September last year, based at its Octillo campus in Chandler, Arizona.

It’s no secret that server hardware updates can be onerous, expensive and disruptive. When your small business has a small IT team and limited resources, it’s tempting to hold off for as long as possible and do your best to maintain the hardware you’ve got. Yet often the smarter choice is to grasp the nettle, find the budget and push ahead. A hardware update can help you reduce long-term costs and risks, but also bring new opportunities to rethink what your servers and supporting infrastructure can do.

In terms of costs and risks, the arguments are simple. Firstly, as servers age, they become more expensive to maintain and run. IDC studies have shown that older servers with lower performance and higher failure rates can increase unplanned downtime by 20%, while server performance degrades by an average 14% per annum, meaning that, after five years of life, it’s 40% slower than when it came out of the box. Older servers aren’t just more expensive to maintain, but also less energy efficient. They may run warmer, needing additional cooling in the server room. This also drives up total cost of ownership (TCO).

Old, inefficient servers can eat up IT resources which could be better used elsewhere; a problem for small businesses, where limited budgets and manpower are the norm. They can also cause issues for performance, stability and availability that impact productivity elsewhere. What’s more, as they reach (or go beyond) their warranty period and end of life, further risks emerge. Software or firmware may no longer be patched to cover vulnerabilities or meet emerging threats. There may be compatibility or performance issues with new applications that your business wants to run.

Take a piecemeal approach, replacing hardware only when it goes beyond repair, and you end up managing a more complex, mixed estate with different hardware and different management tools. Again, this can be daunting for small businesses working to manage that estate without access to a wide range of specialist expertise.

Updating your servers with new hardware can help you minimise these inefficiencies and risks. What’s more interesting, though, is that it could also create new opportunities and help you control your operating costs.

For a start, new server CPUs offer dramatic performance and efficiency gains. AMD’s third generation ‘Milan’ EPYC processors brought AMD’s powerhouse Zen 3 cores to the server marketplace, not to mention improvements in cache, memory bandwidth and IO throughput. These processors offer an up to 2.2x performance increase over the second generation EPYC CPUs, and the performance delta over previous generation Intel Xeon processors is even larger; your existing servers may have CPUs from several generations ago. The range spans everything from single-socket systems with eight-core processors to dual socket systems and 64-core CPUs. Whether you’re a small business with modest but growing requirements or a high-tech startup hungry for processing power, EPYC CPUs have the performance, features and scalability you need.

Of course, it’s not just the CPU that drives performance gains, but also improvements to the platform as a whole. If your existing servers rely solely on magnetic hard drives for storage, new hardware that uses M.2 SSDs for primary storage will be significantly faster. This goes double with the PCIe 4.0 SSDs supported by AMD’s architecture. You may also have access to higher RAM speeds of up to 3200MHz DDR4, and even higher quantities of RAM than your current hardware can support.

It’s also worth mentioning that AMD doesn’t limit its strongest feature set to its two socket server platforms. If you’re operating on a budget, you can take on a single-socket server and still gain all the benefits of the third generation EPYC architecture. Beyond the obvious drop in purchase costs, this could potentially save SMBs a fortune in software licences. When budgets are tight, even smaller savings count.

The point of the exercise isn’t just to run existing workloads faster, but to allow you to do more within a small estate. New platforms with more cores per CPU could host more containers and VMs, giving you the ability to serve more users with more applications, scale upwards when you need to - or even consolidate the workloads you already run. With more cores per CPU, more cache and more RAM, you could virtualise more servers and do more work on fewer physical boxes, using less energy for power and cooling, in less space.

In tests, AMD has demonstrated ten AMD EPYC 7713 servers handling the same workload as 15 servers running the Intel Xeon 8380, consuming an estimated 32% less energy and with an estimated 42% lower TCO. The Xeon 8380 is a recent, third-generation Intel chip. Replacing servers based on older hardware, it’s not unheard of to see organisations reducing their existing server count by up to 75%. That’s a huge plus for sustainability, not to mention TCO.

Virtualisation doesn’t just equal consolidation. With more performance and more headroom for virtualisation, you’ve got the scope to scale up applications on demand, or pilot new applications without splashing out on additional hardware, as and when you need to. There may be new applications in data analytics, machine learning or AI that could bring value to the business, but that your existing servers won’t run.

Plus, by embracing hyperconverged infrastructure, businesses gain the flexibility to manage computer and storage resources across multiple servers in a cluster, making it easier to scale out workloads and use automation. With HCI, even smaller IT teams can run a powerful infrastructure with the capability to do much more than their existing servers ever could. Indeed, vendors can deliver certified HCI configurations designed and built specifically for this job. With new tools and management features, IT teams can manage their estates from a single dashboard, and free up resources for more business-focused tasks.

Updated hardware is also more secure. Not only can you look forward to years of patches and firmware updates, but also new security features. The growth of CPU exploits such as Spectre, Meltdown and Foreshadow has shown CPU security to be critical, and some older architectures to be vulnerable. Newer architectures are designed to resist these attacks, while incorporating additional protections to meet new threats.

All AMD EPYC CPU models feature AMD Infinity Guard technology to secure data in system memory DIMMs, CPU registers and cache. Secure Encrypted Virtualisation (SEV) technology, which encrypts each virtual machine running on the CPU with a unique encryption key, is currently exclusive to AMD. This encryption happens on the fly and with minimal-to-no performance degradation. Supported by many popular hypervisors, it allows your existing applications to run unmodified inside secure VMs. Meanwhile, to ensure system integrity, the server firmware is also protected by AMD Secure-Root-of-Trust technology, to prevent malicious code from booting at startup. You’re not going to find this level of security on older server CPUs.

Perhaps most crucially, updating your hardware can give you a new competitive advantage – the chance to embrace a modern, agile server infrastructure that can power new applications. These applications can help the business find new customers, deliver stronger, more personalised produces and services and reduce product development lifecycles and time to market. They can help them do more with smaller teams and limited resources. Older architectures were designed for the old IT. Modern architectures are ready for what’s next. In short, don’t look at hardware updates as an unfortunate but necessary exercise, but as an opportunity to rethink your IT and make sure your business has access to the tools it needs to thrive and grow.

For a clear idea of how new hardware could help you manage your estate and reduce your TCO, take a look at AMD’s EPYC server virtualisation TCO estimation tool

Intel is set to produce chips for MediaTek using Intel Foundry Services’ (IFS) advanced process technologies.

The agreement is one of the most significant deals the US chip giant has managed to secure since launching IFS last year, in a business space where it competes against the likes of TSMC and Samsung.

The agreement is set to help MediaTek build a more balanced, resilient supply chain through the addition of a new foundry partner with significant capacity in the United States and Europe, the two companies said today while announcing the partnership.

MediaTek plans to use Intel process technologies to manufacture multiple chips for a range of smart edge devices. IFS offers a broad manufacturing platform to do this, with technologies optimised for high performance and low power.

“MediaTek has long adopted a multi-sourcing strategy,” said NS Tsai, corporate senior vice president of Platform Technology & Manufacturing Operations at MediaTek. “We have an existing 5G data card business partnership with Intel, and now extend our relationship to manufacturing smart edge devices through Intel Foundry Services. With its commitment to major capacity expansions, IFS provides value to MediaTek as we seek to create a more diversified supply chain. We look forward to building a long-term partnership to serve the fast-growing demand for our products from customers across the globe.”

The partnership is set to be one of the most significant manufacturing deals Intel has secured since launching IFS last year. The company hoped that IFS would make it a major provider of foundry capacity in the US and Europe, by making custom chips for tech firms and governments and competing with TSMC and Samsung.

The firm’s plan to build its foundry business also aims to address the surging demand for semiconductor manufacturing which was squeezed over the past few years due to a supply shortage.

Intel’s foundry business has already secured other major deals before this MediaTek one, including producing chips for Qualcomm and Amazon last year. This is part of the US chip giant’s plan to regain its lead in the chip-making world by 2025, including by rolling out new chipmaking technologies over the next three years.

Samsung has unveiled its 16 Gb Graphics Double Data Rate 6 (GDDR6) DRAM boasting 24Gbps processing speeds, an industry-first.

Designed to amp up next-generation graphics cards’ performance, the new memory module allows for 30% faster speeds over the firm’s previous 18Gbps offering.

Fitted in a High-K Metal Gate chassis, Samsung’s 24Gbps GDDR6 DRAM lessens current leakage and is fully compatible with all GPU designs. The JEDEC-compliant GDDR6 DRAM is also suited for artificial intelligence-based applications and high-performance computing systems.

Furthermore, upon integration with a premium graphics card, the memory module can transfer up to 1.1TB of data, an equivalent of about 275 Full HD movies, in one second.

"The explosion of data now being driven by AI and the metaverse is pushing the need for greater graphics capabilities that can process massive data sets simultaneously, at extremely high speeds," said Daniel Lee, executive vice president of the memory product planning team at Samsung Electronics.

"With our industry-first 24Gbps GDDR6 now sampling, we look forward to validating the graphics DRAM on next-generation GPU platforms to bring it to market in time to meet an onslaught of new demand," added Lee.

Samsung has already started sampling its 24Gbps GDDR6 DRAM, with plans to commercialise the memory module in tandem with the firm’s GPU platform launches.

In February last year, Samsung debuted the industry's first artificial intelligence (AI)-powered memory unit, with the architecture combining high-bandwidth memory with AI processing power to boost performance and lower power consumption.

Intel has started informing customers of a planned price hike for products in various market segments, while internal sources have reportedly revealed the prices will begin to rise in Autumn 2022 due to the rising costs of production.

The company told IT Pro its plan to raise prices was previously announced in its Q1 2022 earnings call but further details on the products and business segments that will see the hike are yet to be revealed.

According to sources speaking to Nikkei Asia, the products included in the price rise will include those that are core to its PC and server business, in addition to other areas such as the chips required for networking products.

Intel told IT Pro that it has not publicly disclosed the specifics of the price changes, nor has it publicly disclosed the impacted products. However, one source speaking to Nikkei Asia said the percentage increases could be anywhere between the single-digit range and maybe even more than 20%.

A similar price rise was also announced by TSMC, the world’s largest contract chipmaker, saying it planned to hike prices similarly by up to 20%.

The increases may also vary depending on the type of chip. When IT Pro approached Intel for confirmation of the reports, the company said: "On its Q1 earnings call, Intel indicated that it would increase pricing in certain segments of the business due to inflationary pressures. The company has begun to inform customers of these changes".

The semiconductor shortage in the technology industry is well documented and has persisted since 2020 after a range of different factors, such as geopolitical tensions, pandemic-related shifts in demand, and disruptions at foundries across the globe, among other factors, all culminated in a global shortage.

Most businesses that rely on semiconductors have felt the effects of the scarcity, including the GPU market that has famously impacted the video game industry and PC builders.

Cisco is another major company that has experienced financial hardship as a result of the shortage, with the company thought to be sitting on a backlog of orders that it currently can’t fulfil worth billions of dollars.

Most nations around the world are now facing rising inflation rates, in part due to the ongoing war between Ukraine and Russia, combined with the lingering pandemic-related hits to their economies.

It was revealed this week that both the US and UK's inflation rates hit 40-year highs of 9.1%, while the average for countries with advanced economies, members of the Organisation for Economic Cooperation and Development (OECD), has at least doubled since 2020.

The UK is facing a cost of living crisis as consumer prices in areas such as utilities have soared in recent months, and will continue to soar into the new year, according to forecasts.

A quarter of UK businesses are also expecting to raise their prices as a result of rising inflation rates, according ONS figures.

Intel admitted in its most recent earnings report that “the pandemic has significantly increased economic and demand uncertainty,” and that the “demand for [its] products could be materially harmed in the future” as a result.

Questions around when the chip shortage will end are currently met with mixed predictions and Intel had previously warned that it could extend well into 2023.

Intel and Google Cloud have co-developed a new toolkit to simplify and optimize high-performance computing (HPC) workloads on Google Cloud.

Dubbed Cloud HPC Toolkit, the new platform allows enterprise customers to create and replicate turn-key HPC clusters. Backed by Intel’s 2nd Gen Xeon Scalable processors, the solution also features the firm’s Select Solutions.

Additionally, the new resource provides access to tools from “Intel oneAPI Base and HPC Toolkits to optimize performance through Intel Select Solutions for Simulations & Modeling.”

“Using Cloud HPC Toolkit with an Intel Select Solutions for Simulations & Modeling blueprint brings the added benefit of automatically spinning up a hardware-software configuration that has been rigorously tested and optimized for real-world performance, eliminating guesswork,” stated Ilias Katsardis, HPC solution lead at Google.

Furthermore, built-in Intel oneAPI tools streamline HPC programming by offering support for C++, SYCL, Fortran, OpenMP, MPI, and Python. Commenting on integrations, Google affirmed support for Altair PBS Professional and Altair Grid Engine job scheduler in later versions.

Intel had previously partnered with Google to automate the creation of Intel Select Solution-compliant HPC clusters on Google Cloud. The recent announcement reinforces the duo’s longstanding partnership.

“DAOS is the future of HPC. We’re excited to see the benefits of our year plus technical collaboration pay off with today’s announcement of fast and easy access to DAOS in Google Cloud,” commented Kelsey Prantis, director of engineering, high-performance storage, at Intel.

Laptops are now essential to modern business. They offer flexibility, mobility and high end functionality which enables employees to meet their organisation's needs and deliver on key objectives; in fact, it's all but impossible to imagine what a modern business would look like without access to notebook PCs.

Their significance has become even more acute since the coronavirus pandemic led to an enormous surge in remote working. Without laptops, many businesses would have been unable to continue operating during the extended lockdown periods, and even now that restrictions have lifted, remote working is firmly embedded as an accepted norm.

This surge in remote working has fed into a growing desire from businesses of all stripes to pursue digital transformation and innovation in order to establish a competitive edge, but the responsibility for enabling these technical capabilities is often placed in the hands of IT teams, with little extra resources (if any) to support this increased scope. This added burden has put more pressure than ever on the time and workloads of technicians.

Streamlining and time-saving

One of the biggest challenges for any IT team is simply keeping the wheels turning and the lights on, ensuring that all an organisation's infrastructure and equipment is functioning normally. However, while the regular monitoring and maintenance of IT systems is essential work for keeping a business’s technology operational, the more time that can be freed up from performing these tasks – which are often repeated, manual processes – is time that can be spent on enacting some of the more ambitious transformation initiatives that may deliver increased value overall.

This is why enterprise device management solutions such as the tools built into the AMD PRO platform have become one of the cornerstones of business IT. They allow organisations to administrate all of the laptops in their fleet from a single, centralised dashboard, quickly identify problems and begin working on remediation efforts. This helps speed up workflows and increase efficiency, allowing smaller teams to manage larger numbers of employee devices. With technical talent in such high demand, this means that the growth of your business won’t be held back by the growth of your technical support staff.

When a new Windows update is released by Microsoft, for example, IT admins can automatically roll it out to all devices in the organisation on a pre-set schedule to minimise disruption, while ensuring a speedy upgrade without having to install it machine by machine. Rolling out upgrades manually on an individual basis is a long, tedious process, and the ability to take care of the update process systematically using native tooling frees up large amounts of IT time while granting access to the latest features and security improvements. Similarly, if for any reason the update needs to be rolled back, this can also be quickly and easily accomplished without having to physically recall devices.

This means that issues can be resolved much faster, which is particularly important for fast-paced modern businesses. After all, downtime can be expensive, and in most organisations, an employee without access to their laptop is an employee that has little means of being productive. Enterprise manageability tools help IT teams monitor employee devices so they can proactively fix a range of potential hardware and software problems before they turn into major issues. They include the ability to remotely change settings, apply software fixes and even trigger system restores to roll the device back to a previous stable recovery point.

Secure in the knowledge

Asset inventory is another key capability. Knowing exactly what devices are in your fleet – as well as their hardware configuration and firmware status – is a vital part of maintaining a robust security strategy, and the ability to invisibly gather this data and centrally log it without having to manually audit devices is one of the principle reasons why enterprise manageability tools are so important. In addition, they also help assist with compliance practices, and allow for more efficient coordination of warranties, support contracts and device refresh plans.

Because AMD PRO’s management tools are designed to integrate seamlessly with Microsoft Endpoint Manager, organisations can leverage these abilities without having to adapt their existing device management stack to accommodate them. what's more, because they're based on open standards and DASH profiles, there's no risk of vendor lock-in, and a comprehensive range of third-party enterprise mobility management solutions are also supported to ensure that IT teams can manage their AMD PRO-equipped fleets as easily as possible alongside the rest of their devices with a consistent experience.

All of these capabilities are available remotely too, which is essential for organisations that need to support workers both at home and in the office. Hybrid working is now a fact of life for the majority of businesses, and having the right tools to address problems from anywhere ensures not just that staff can continue working without having to grapple with technical hiccups, but also ensures they feel happy and looked after, which translates to better employee satisfaction and reduced turnover.

Alongside the robust platform validation, long-term software stability and multi-layered hardware-based security protections, these features have been designed to make AMD PRO a processor platform built for the needs of modern, agile business computing.

Read AMD’s ‘The challenge: Doing more with less’ to learn more about enterprise manageability

Just as there’s a direct link between the performance of a business and the performance of its employees, so there’s a link between employee productivity and the hardware they use on the job. Slow application performance, unresponsive software or an unreliable PC can all have consequences. A 2020 UK study found that nearly half of British workers felt that outdated tech hindered their productivity, and that slow technology cost them, on average, up to 46 minutes a day. With a speedy, responsive PC, however, users have time to focus and work creatively, rather than wonder how long they’re going to have to wait while their PC grinds its gears.

This makes it crucial that IT teams refresh their PC and laptop fleets regularly. Yet this raises another issue: How do you evaluate performance and make meaningful comparisons that can guide you towards the right choice?

Evaluating PC performance has rarely been simple or straightforward, particularly when it comes to CPUs. Back in the 1990s you could make some assumptions based on clock speeds; the more MHz you had to play with, the faster your applications ran. The problem is that while many businesses still make these assumptions, maybe factoring in core counts, threads and – possibly – cache, these methods of assessment are fundamentally flawed.

As a new white paper from AMD, ‘Measuring what matters – Benchmark considerations for commercial PC purchases’, makes clear, modern processor design is less focused on ramping up clock speeds, and more focused on creating efficient super-scaler designs with deep instruction pipelines and a more sophisticated use of cache, enabling each core to execute multiple instructions per clock cycle. In fact, designers may even focus on reducing the number of clock cycles required to execute a given set of instructions, rather than rely on pushing clock speeds upwards. If you automatically assume that a processor running at 3 to 4.8GHz is going to give you more performance than one running at 1.9 to 4.4GHz, you could be making a big mistake.

In fact, AMD’s white paper suggests, “The idea of boost and base frequencies is an oversimplification – the actual frequency at any given time will depend on workload, power-mode and platform thermal design.” This is true of desktop PCs, but even more so of laptops, where the thermal design of the device will constrain how long the CPU will be able to run at its maximum boost frequency. In real world usage, one CPU in one laptop might reach a higher boost frequency, but quickly throttle to a lower speed as power demands increase and temperatures rise. Another CPU in another laptop might reach a lower boost frequency but prove able to sustain it over longer periods of time.

Meanwhile, core counts matter. Provided an application is designed to make effective use of multi-threading – and most are these days – the AMD CPU with eight cores running 16 threads will still run a given operation faster than the Intel CPU running four cores with eight threads, higher clock speeds notwithstanding. In fact, it might find it easier to stay within thermal constraints and sustain higher clock speeds longer. Throw in all the other variables within the system – memory bandwidth, storage bandwidth, GPU performance – and simple MHz comparisons can be misleading.

So, how can businesses get a more accurate idea of real-world PC performance? The ideal would be a programme of user testing, with tests designed to mimic everyday real-world tasks in real working environments. The problem is that this is long, difficult to manage and potentially expensive, while it's difficult to pull out consistent and unbiased metrics. Software benchmarks can be more practical, but they need to be approached in the right way.

The obvious answer is application benchmarks, either using industry-standard PC benchmarks or, where the development resources are available, custom scripts based around real-world business tasks. However, both approaches have their issues. With industry-standard benchmarks, businesses need to make sure that they’re a good fit for the organisation, and not focused on consumer workloads or creative workloads that don’t reflect the kind of applications they actually use. Custom benchmarks, meanwhile, need regular updating to ensure that they’re not optimised for older architectures and still provide accurate, relevant results.

Both types share the problem that, while they can provide an accurate picture of performance in existing applications, they can’t predict how PCs will perform with new or emerging applications. What happens when a new, revolutionary tool appears that uses specific hardware features, but your benchmarks can’t tell you if a PC measures up?

This is where synthetic benchmarks come in. They can provide a snapshot of system or component-level performance, giving you some idea of how system X compares to system Y, or the read/write speeds of one SSD when compared to another. They can be quick, dirty and effective, but they need to be used with caution. As AMD’s white paper notes, narrow benchmarks of, say, the storage or memory subsystems may not call on features in the CPU pipeline or cache, skewing the result. “It is important not to overly weight these results,” it concludes, suggesting that “decision makers should instead evaluate the composite score, which exercises a broader set of processor functions”.

What’s more, synthetic benchmarks may not run for long enough or put enough stress on every part of the system to assess how performance will hold up against issues with memory bandwidth, I/O latency or thermal constraints.

It’s clear that no single benchmark or type of benchmark works for evaluating PC performance on its own, but where does that leave us? AMD’s white paper takes a smart approach, which combines “a wide range of both application-based and synthetic tests”. Through this kind of holistic, comprehensive testing, AMD shows that the Ryzen 7 Pro 5850U processor exceeds its rivals in a range of real-world applications, even though it runs at lower clock speeds and falls behind in some synthetic, single-threaded tests. No single benchmark can tell the whole story – and MHz to MHz comparisons won’t even get you past ‘once upon a time’. If you’re serious about evaluating PC performance, you really need the wider picture.

Read AMD’s ‘Measuring what matters – Benchmark considerations for commercial PC purchases’

Over the past few years, countries around the world have suffered from an acute semiconductor shortage that’s had a fundamental impact on global trade and supply chains. World leaders have slowly begun to understand the importance of an uninterrupted chip supply, having borne witness to the way their domestic industries have been hit.

The crisis has spurned many countries into attracting global chipmakers into their territories to avoid future supply chain disruptions. This is especially true as technology becomes increasingly integral to the global economy.

India is no different, and has more of a reason than others to attract such industry giants. This is because it’s hoping to emerge as a growing tech hub, evidenced by its focus on this sector in its latest budget. Gartner forecasts government IT spending to total $8.3 billion in 2022 – an increase of 8.6% from 2021. Due to this growing status, India has embarked on an aggressive programme to attract companies to establish fabrication plants – or fabs – as it strives to become a global semiconductor manufacturing hub.

India’s growing manufacturing hub

Because it’s evolving as a growing manufacturer for consumer appliances, automotive, and industrial products, India hopes to attract chipmakers rather than continuing to import.

In December 2021, the government approved the SemiconIndia project which set aside $10 billion for the development of a sustainable semiconductor and display manufacturing ecosystem. The government, in February, then invited companies to take part and submit applications to establish semiconductor and display fabs across the country. Ultimately, it received $20.5 billion in investment proposals from five companies to accomplish this.

As part of this silicon rush, three applications were put forward: Singapore’s IGSS Ventures, mining and metals group Vendeta in partnership with Foxconn, and Israeli chipmaker ISMC. The companies are hoping to invest a total of $13.6 billion into semiconductor fabs and are seeking around $5.6 billion in government support.

Then, in April, prime minister Narendra Modi underlined the urgency by repeating his rallying cry for further investment into India’s growing semiconductor manufacturing sector, as he set his sights on Taiwan and the US.

Modi (pictured) is pursuing an aggressive programme to build a semiconductor manufacturing industry from scratch

Despite the government’s multi-billion dollar commitment, India still currently doesn’t have any large-scale semiconductor fabrication facilities of its own in the country, says Kanishka Chauhan, principal analyst at Gartner. “This is an ideal opportunity, as, if the fabs are set up, it can potentially fulfil some of that domestic demand, reducing dependence on other regions,” says Chauhan.

There’s now a much greater emphasis on self-reliance throughout the world, partly due to global events like the US-China trade war, COVID-19 and Russia’s invasion of Ukraine. India isn’t alone in this shift, however, with many countries aiming to reduce their dependence on each other. The EU proposed a €43 billion Chips Act to tackle semiconductor shortage, for example, while TSMC and Japan partnered to build a $7 billion chip plant in Japan.

Sampath Kumar Venkataswamy, senior research manager of Manufacturing Insights at IDC Asia/Pacific, agrees, also citing other events like the drought in Taiwan, fleet delays, and increased transportation costs. These have led to several organisations either delaying their productions plans or cancelling them.

“Given the fact that chips are an essential part of the technology driven product value chains, the Indian government is aiming to address those by bringing manufacturing in-house,” he underlines. “The other benefits include increased innovation, improved margins, increased GDP contribution and a business model that can further scale and support products beyond smartphones.”

Hurdles in establishing a chipmaking presence

The global direction on travel on self-reliance is fuelled by semiconductors paving the way for the future, says Chauhan. They’re used in everything from artificial intelligence (AI) systems to cloud computing and autonomous vehicles. Despite this, it’s difficult for chipmakers to establish themselves in India. A chip manufacturing plant can be a costly investment, ranging from between $5 billion and $20 billion, depending on the type of plant.

If a company decides to invest, however, Chauhan underlines they need to be assured of a few things. The first is the market, because if there’s no market for the products, they will become meaningless to produce. Supplying to other countries may not be an ideal choice as logistics could complicate this.

Next, the return on investment can take years, and it becomes the government’s responsibility to assure the companies they won’t face any challenges, or surprises, when it comes to taxation or regulations.

Chipmakers also need to be assured that local resources are available. “Chip manufacturing is a resource-intensive process,” explains Chauhan. “An uninterrupted supply of water and electricity is needed. Even a momentary disruption in the power supply can cause major issues in the manufacturing line causing time and money.”

Additionally, the plants will need access to skilled labour. Since the industry is starting from scratch, finding the right talent can be a huge challenge. Chipmakers may have to think about instigating the training and recruitment of talent before they set up shop.

Lastly, the fabs won’t be able to survive without the presence of a chip ecosystem. This means vital constituents of this ecosystem need to be encouraged and supported, including chip designing, outsourced semiconductor assembly and test (OSAT) or assembly testing marking and packaging (ATMP).

“I think there are very high chances that this initiative is going to gather more momentum,” says Chauhan. “The international semiconductor consortium ISMC has already signed an MoU with the government. However, what companies finally set up and what they make, will dictate the outcome of this initiative. Setting up the plant is not enough.”

Combatting the chip talent shortage

While India produces a large number of skilled workers, the presence of semiconductor facilities within the country is scarce in comparison to others, says David Stone, chief executive of MRL Consulting Group.

This leaves the country reliant on importing chips and forces a huge amount of the country’s skilled population to leave in pursuit of a career in the semiconductor industry. “The current semiconductor shortage disrupted supply chains, and COVID-19 revealed how vulnerable a country can be if they don't produce semiconductors themselves,” adds Stone.

The talent in the country is also heavily skewed towards software and fabless design. One of the biggest challenges organisations will have to deal with is the availability of skill sets that are aimed at chip manufacturing and refining. “Embedded chip design and integrations has typically been India's strongholds, with several large companies such as Intel and Texas Instruments holding Indian captives,” believes Venkataswamy.

India’s national and local governments are making strides towards making their country a desirable place for plants to be established, says Stone, referencing the multi-billion dollar commitment. “However, the past two years of unstable supply chains, scarce raw materials, and huge fluctuations in energy prices have proved how difficult it is to maintain and operate fabs in even the most established of countries,” adds Stone. He says India has always faced issues with stable and reliable access to energy, clean water, and fuel, and is prone to natural disasters which will threaten any chipmaker in the country.

India plays host to a steady stream of tech talent

The good news is India produces a huge amount of chipmaking talent. This is as much as 20% of the global population of design engineers, according to the prime minister. The Chips to Startup ecosystem, announced in December 2021, also promises to produce a further 85,000 engineers, and helps to maintain the country’s local talent pipeline. Bangalore, for example, has a high density of skilled workers and houses over 50 semiconductor companies, making it the country’s most prominent tech hub, says Stone.

“India has a huge population of skilled workers, and the government has seemingly recognised the importance of semiconductors, providing their workforce with the infrastructure to stay within their country and grow their global presence in the chips industry,” says Stone. “If India is able to solidify reliable access to energy and clean water, it will provide a fantastic environment for any chip maker to set up shop.”

Gaining chip independence

Most of India’s electronics industry is centred around assembly and packing, which doesn’t yield much benefit for companies looking to add value to products and services, and increase their revenue and bottom line, says Venkataswamy. The government wants to fundamentally boost the economy, which is only possible by making fundamental components like microchips, he adds.

SemiconIndia, among other schemes, will help to attract global players. The basic requirements around the consistent availability of power and water to support chip manufacturing operations, though, still needs to be addressed.

“The ease of setting up a business in India has improved over the years and the red tape associated with it also has been simplified substantially, but the challenges associated with early-stage business planning and growth still apply,” says Venkataswamy. “Established chip makers will find difficulties in sourcing operations because of a nascent ecosystem which still relies heavily on imports.”

With other nations across the globe also looking to attract chipmakers, such as China, India has its work cut out to convince these organisations it’s serious about becoming a global competitor. It’ll need to wean itself off imports and also ensure it has the uninterrupted supply of energy and water that fabrication plants require. This could be tricky, especially considering how climate change is impacting the country and the way it produces energy.

However, if there’s a time for India to seek investment into its semiconductor manufacturing sector, that time is now. After years of supply chain pressures, it makes sense for the country to invest in its chipmaking capabilities, kickstart the industry and reap the rewards.

Many frustrations and failures in business computing come from things that ought to be simple. The problem is, it’s easy to focus on big, showy, chewy problems and neglect the little things that can have a big effect on overall productivity. With a little planning – and sometimes a little user participation – there’s an awful lot you can achieve in a single day, with minimal disruption to the business and no need to book in expensive consultants or advisers.

Aside from the direct object of a project, there can be a cultural benefit too. IT departments are largely reactive, but a one-day project gives them a chance to demonstrate a bit of proactivity and forward thinking, without completely rewriting their role. Involving the users encourages a bit of self-reliance too – an increasingly desirable trait as hybrid work becomes the norm. Ultimately, the odd one-day project can only be positive for the relationship between IT staff and the rest of the business. Here are some ideas to start you off.

Fire drill day

This is the granddaddy of one-day projects – when you take an afternoon out to pretend there are flames shooting out of the back of the server and spreading around the building. The experience can be both hilarious and chastening, with lessons that go far beyond the IT department as it sinks in exactly how reliant the business is on its flammable assets.

Indeed, in these days of cyber resilience and perilous international relations, it’s worth considering extending the fire drill concept to specific bits of your IT environment. Take a day out to thoroughly investigate the failures and shortfalls that take place when a particular system goes offline and can’t simply be brought back with a remote power cycle.

Log file reading day

I can see the wrinkled noses as I type. Log files? What use are they on a laptop? Actually, they’re just as much use as they were on the Windows 2000 Server where you first saw them. It’s true that most Windows PCs go for years without the Event Viewer ever being opened, but that’s not something an IT department should be proud of – the information contained therein can provide answers to everything from obscure network problems to pinpointing the exact moment power was restored in one of those sneaky overnight outages we’re not supposed to fret over.

It doesn’t require much specialist knowledge, either. Events aren’t mystical things; if anything they’re a bit too explanatory, making the beleaguered sysadmin work laboriously through the verbiage to get to the crucial information. But at the end of the hunt, you’ll often find references to illuminating knowledge-base articles. And you may start seeing patterns in distinct devices: persistent loss of connection to shared network drives? Exchange servers you can’t reconnect to after your machine’s gone to sleep? Such subtle failures become much more obvious when you let the Event Viewer do the searching.

There are a slew of management utilities which will consolidate many machine’s events and log files and go looking for patterns in the mash of data thus created. However, quite a lot of them are integrated into much larger network-management software suites, which with the best will in the world, you’re not going to deploy as a one-day project.

Hunt the temp files day

While I might not want unqualified hands wandering about the file structure of my Exchange server, there’s definitely no harm in getting a bit of user-participation going when it comes to finding and nuking the myriad types and collections of files which you can’t see are being collected by your PC.

Don’t underestimate the potential gains. Some applications generate huge temporary files; others churn out smaller items, but in great numbers. I’ve seen printers create a single file per print job or sheet of paper, and not always cleanly delete them once the print run is over.

It’s not just about disk space. There are a whole raft of different methods of picking up a file in a directory, from the perspective of a developer, and temporary files can have a big impact on performance. This is why things run faster in neglected Windows 7 machines once you clean out all the .txt files in the Windows directory, and why periodically emptying C:\WINDOWS\Temp can seem to turbo-charge certain operations.

There are downloadable utilities which can automate the process of clearing out temporary files, but it doesn’t hurt to engage your staff and help them gain some understanding of these not-too-mysterious secrets of business computers in action.

The other Patch Tuesday: Physical cabling

Are you sure that all the leads and interlinks that make up your physical network are working as they should? Really sure? After all, even the small patch leads used in an equipment rack can get pinched by heavy servers and steel uprights.

This isn’t something you can do in software. Faced with a big pile of patch leads, your best bet might be to employ a Fluke cable tester, though those come with a reassuringly plump price tag. However, many modern switches can report on cable damage via their web interfaces, and there are cable-testing modes hidden away in some of Intel’s network card drivers too. Only the most hardcore techies tend to discover those, not least because they don’t show up when you’re using the default Microsoft drivers.

Checking your cabling makes for an excellent one-day project, because if it gets to 4.59pm and you haven’t quite finished, you should still be left with a stack of known-good cables. Admittedly it’s quite a boring process and not one to leave with the users, but it shouldn’t get in the way of normal operations either – briefly pulling a cable out of a PC and replacing it barely interrupts the session, and both ends ought to resume the link seamlessly.

Drivers and firmware day

I’ve suggested involving non-IT staff in your one-day projects, but this is one I’d recommend you leave to the specialists. Playing with firmware used to be a reasonably safe proposition, because the sorts of device that had firmware used to be ones that did quite simple jobs. Now, though, the firmware has become more complex and the devices themselves have become more critical: even if your router isn’t running the latest firmware, you can’t just shut it down for an update.

To be honest, when dealing with new clients and networks in unknown states of repair, I always try to see whether, rather than performing an in-place upgrade to a vital core bit of the network, we can replace it with something newer and better. That doesn’t always mean consigning that old device to the Cupboard of Shame: quite often it turns out that running two routers in a lightly connected setup gives you an overall faster design, while mitigating the risks of firmware upgrading.

Of course, Wi-Fi routers aren’t the only devices with firmware. Even the humble network switch has a brain, and it’s worth making sure it’s up to date in the name of security and performance. Then, of course, we get into the wonderful world of user hardware: it’s quite incredible how much better the latest Intel Ethernet drivers are, compared to the ones burned into the initial releases of Windows 10. Unless you’re working in a very small office, however, it might be a stretch to get the drivers found, identified, downloaded and matched to all your PCs in a single day. And I’m willing to bet a small pile of actual coins of the realm that even if you can do that, you won’t have time to go round and optimally configure all the extra network parameters and controls provided by the driver.

‘Wooden Spoon’ discovery day

Businesses grow in fits and starts. It’s very unlikely indeed that all the PCs in your office are of the same age, spec, and operating system revision. I’d go so far as to say that the only companies where I’ve seen such standardisation are the ones in which I’ve recommended it, to provide a bit of resilience and redundancy when things start going peculiar.

The Wooden Spoon is of course the award that goes to the oldest, slowest, most decrepit computer that’s still in regular use. For logical but ironic reasons, this will frequently be the one that’s doing the most business-critical job, and being used by the most hard-pressed member of the firm. If you can prise them off it for just a day, the gains can be enormous: I once replaced a Pentium 4 workstation with something that had two six-core Xeons and an onboard RAID card, representing a leap of about six hardware generations in one afternoon. The old machine had been taking about 40 minutes to run the developer’s modelling code; the new one ran the same code so fast we couldn’t press the stopwatch quickly enough.

There’s no need to stop at workstations: printers, network switches and other key components of your business have a tendency to quietly reach retirement age while your attention is focused elsewhere. Tread a little carefully though. You can’t always just whisk away an outmoded item and have a fully functional replacement configured and running by the end of that singular day. Start with a focus on identifying the worst mismatches between hardware and job requirements, and make a realistic plan from there.

Password day

Managing access – to cloud systems, websites, customers and all sorts of other things – has never been more important. Yet the actual management of passwords has become increasingly fluid. In practice, almost every worker nowadays has two sets of passwords – their own, and all the other ones they need to use in the course of doing their job. Password reuse has lamentably become so commonplace.

It’s vitally important, therefore, to find out what passwords everyone’s actually using – along with where they got them from, and who else knows them. Obviously this isn’t something that should be on a cloud-hosted spreadsheet, though: locking a password booklet in your office desk drawer might be smart. It might then even survive a fire drill.

Shadow amnesty day

“Shadow” here refers to shadow IT – the unsanctioned use of third-party services inside a business. It might be Google services, Skype or what have you; plenty of workers, especially the younger generations, are wholly accustomed to using free consumer tools to get things done, and won’t hesitate to fire them up in a professional setting. It could also include any sort of activity that goes outside of your documented and managed processes counts.

The aim of a shadow amnesty day isn’t to cut out all the shadow service uses. Primarily, it’s to identify the areas where your in-house, properly managed IT department aren’t giving users what they need. If you can then use that information to set up a proper sharing platform that means people no longer need to put confidential documents on Dropbox, that’s a win for all concerned.

Advanced project: Duplicates day

This one isn’t all that advanced for techies, but it can feel that way for users when you ask them to go hunting around their systems for duplicate files. Whether you use Windows 10 or Windows 11, the operating system doesn’t make this easy, especially when documents are spread across a variety of storage platforms and security settings.

You might question whether this is a worthwhile use of time and energy. Many people do – many people, that is, who haven’t seen the benefits of de-duplication, as found on the most upmarket NAS and storage server gear. These monsters – a single full-height rack packed with disks qualifies as a small one – can identify data that’s duplicated across multiple files and users, and refer hundreds of thousands of files all to a single block copy. The effect on storage requirements can be breathtaking: I’ve seen the time taken to back up an entire Exchange Server virtual machine (VM) drop from several hours to a matter of minutes.

Okay, so we’re not going to match those benefits by clearing out a load of individual duplicates, but we can still make a significant difference to your storage needs, which will in turn have a knock-on effect on the speed and efficiency of your backups.

The best approach is to standardise on a few lightweight utility tools. The first task is to analyse exactly where your storage is being eaten up, and to identify where you most need to free up some space. Then, you can use one of numerous free tools to search and destroy any space-wasting duplicates you may be harbouring. Personally I split up these operations by size; I start by searching for duplicate files over 500MB and then work my way down. This way, you can achieve big space savings very quickly, and it’s up to you whether you continue to inspect the smaller files.

US chipmaker Qualcomm is seeking to invest in Arm as part of a consortium.

Arm’s owner SoftBank is hoping to list Arm next year after Nvidia’s $66 billion acquisition of the company fell through earlier this year.

Cristiano Amon, Qualcomm’s chief executive, told the Financial Times that the company is an interested party in investing, underlining that Arm is a very important asset given its crucial role in the global technology sector.

Amon emphasised that Qualcomm could join up with other chipmakers to buy Arm if the consortium making the purchase was big enough.

“You’d need to have many companies participating so they have a net effect that Arm is independent,” said Amon.

Qualcomm had opposed Nvidia’s proposed acquisition of Arm, saying that it made no sense for one chipmaker to take control of a company that was of such fundamental value to the entire sector.

“Arm has won everywhere because of the collective investment of the entire ecosystem, from companies like Apple and Qualcomm and many others, and that’s because it was an independent, open architecture that everybody could invest in,” said Amon, talking about the period before SoftBank purchased the company.

Qualcomm hasn’t spoken to SoftBank about the potential investment just yet, revealed Amon, He said the Japanese group had been prioritising resolving a dispute at Arm’s joint venture in China, where its CEO had refused to step down.

Amon’s proposition could provide additional momentum to the idea of a group of chipmakers becoming investors in Arm. Intel CEO Pat Gelsinger suggested the company could support a move earlier this year.

In March, SK Hynix was also reportedly reviewing plans to acquire Arm. Park Jung-ho, vice chairman and CEO of the company, said it was considering forming a consortium, along with strategic partners, to jointly acquire the chipmaker. He emphasised that he doesn’t think Arm can be bought by one organisation.

Arm strikes licensing deals with partners, no matter their size or where they are based. This has led to its intellectual property being used in the majority of chips sold globally. In August 2021, the Competition and Markets Authority (CMA) raised concerns around Nvidia’s acquisition of Arm as the merged business would have the capacity and incentive to restrict access to Arm’s intellectual property. This is used by companies that rival Nvidia to produce semiconductor chips, in competition with the US giant.

This was echoed by the US Federal Trade Commission (FTC) in December 2021. It said the deal would "stifle" competition for multiple technologies because Arm is a "critical input" that fosters competition between Nvidia and its rivals. It said this kind of acquisition would give Nvidia a way to undermine other companies.

Why is Arm so important?

Consumption of semiconductor chips is growing rapidly, in the EU alone the consumption of leading edge chips is forecast to grow exponentially by 15% compound annual growth rate between 2020-2030, Dieter Gerdemann, partner at Kearney, told IT Pro.

"It should come as no surprise that due to the dominance of Arm, in the supply of chips for smart phones, with 90% of devices using the chip and IoT applications, and in fields such as data centres that Qualcomm has chosen to take an interest," he added.

Gerdemann said that Nvidia's attempted takeover of Arm was met with critique across the semiconductor sector which argued for the independence of the ARM eco-system and the benefits ARM’s status as a chip powerhouse in Europe brings. 75% of leading-edge chips are currently being produced in East Asia and, following the EU Chips Act, the focus has never been more on Europe.

“The idea that Qualcomm might now invest in Arm, after arguing against the Nvidia takeover, is an interesting one," he said. "The consortium they indicate to invest through will certainly need to meet stringent requirements for openness and transparency in order to gain acceptance of the global semiconductor community.

"However, we are convinced of the benefits of a publicly listed Arm entity, as the forces of the free market will ensure Arm stays competitive and continues to be the innovation powerhouse that it is today – to the benefit of the semiconductor ecosystem and Europe as a semiconductor design powerhouse.”

Buy a Windows computer today and it’s all but certain to contain a Trusted Platform Module (TPM) – an unassuming component that performs a standard set of security functions. Those functions are defined by the Trusted Computing Group, a non-profit industry body with more than 100 members, including major laptops hardware manufacturers AMD, Dell, Google, Intel, Lenovo and Microsoft, working together to improve security across computing devices and applications.

The group has existed in various forms since 1999, but it was in 2009 that it published the first mainstream TPM standard, curiously presented as version 1.2. This quickly gained widespread adoption across the industry, thanks partly to support built preemptively into Windows Vista. In 2015, the newer TPM 2.0 standard brought some extra features, and this remains the state of the art when it comes to trusted computing.

Despite its ubiquity, the TPM has always been an optional component. It’s been up to computer manufacturers whether they’ve chosen to include one in their designs, and up to users whether they want to enable or disable it in their BIOS settings. With Windows 11, however, the mood has changed: Microsoft has declared that the latest edition of its operating system will only be supported on systems where TPM 2.0 is present and enabled.

That’s been a controversial decision, partly because it seems arbitrary. Windows 10 didn’t require a TPM, and it’s perfectly possible to use Windows 11 without one. Microsoft even provides a simple hack that lets you install its latest operating system (OS) without a TPM. If you take this route however then your computer will be officially classed as unsupported, meaning it’s not guaranteed to receive future updates. So why has Microsoft decided to make TPM 2.0 a requirement? To understand, let’s look more closely at what a TPM is and what it does.

Where is the TPM?

TPM was originally conceived as a small, low-cost chip that could be optionally plugged into a dedicated socket on a motherboard. As TPM support became the norm, manufacturers started soldering the chip directly onto the board.

The specification doesn’t require that the TPM be implemented as a physical chip, however. In fact, nowadays the full set of TPM functions can be integrated into the CPU firmware – as is the case with every processor on the Windows 11 supported hardware list, from both AMD and Intel.

If the Windows 11 installer complains you don’t have a TPM, it might be that the feature simply needs enabling in your BIOS. If you can’t find an option to turn it on, you might need to look for a different name: AMD’s integrated TPM system is called the Firmware Trusted Platform Module (fTPM) while Intel’s is called Platform Trust Technology (PTT).

What does the TPM do?

The TPM works like a miniature computer in its own right. It receives commands, it returns responses, and it can also store and regurgitate information. To enable it to do this, it has its own onboard storage: the specification requires that a TPM must have at least 7KB of storage, but newer modules can have upwards of 100KB. All of this memory is non-volatile, meaning it retains its contents when unpowered, like an SSD.

Of course, if you just want to store and retrieve data you might as well use the hard disk. What makes the TPM special is that stored data can have conditions attached to it, governing how it can be accessed. For example, if an application stores a password in the TPM, it can mark it as only accessible to the same process that originally provided it. If any other programme requests the information – such as a piece of malware trying to steal your secrets – the TPM will refuse to provide it. Since the workings of the TPM are completely separate to and outside of Windows, the malware is stymied.

A TPM can also deny or allow a data request depending on the state of the computer. This is because the TPM has the ability to audit the hardware configuration of the computer it’s installed in, and to log the software that launches on it. Using this information, it can determine whether or not it’s in a known, trusted environment, and respond accordingly.

A different use for the TPM is to enforce the entry of a password when the computer first powers on. The BIOS brings up the password request, but the authentication is handled by the TPM, using credentials stored in its own onboard memory – so you can’t get around the password by clearing or flashing the BIOS.

The TPM even has protections against brute-force attempts to guess the password: after a few wrong entries, the TPM will start to make you wait an increasingly long time until you can try another password. Too many wrong guesses results in an enforced delay of 24 hours – and don’t bother trying to change the date in the BIOS or whip out the CMOS battery, as the TPM has its own onboard clock that only ticks down while the system is powered on.

This is all good stuff: the TPM can help prevent thieves and intruders from accessing protected systems, and block malware from exfiltrating sensitive data. But we haven’t yet come onto perhaps the most important function of a TPM.

Cryptographic magic

The TPM can store data as-is, but it can also perform cryptographic functions on it. Specifically, it can apply public-key cryptography – the type that involves a pair of encryption keys, one of which you keep secret while the other is shared far and wide. Anyone can use your public key to encode a message to you, but only your private key can be used to decode it. Or, your private key can be used to generate a cryptographic hash of a message or file in a way that can be validated using your public key, proving it was “signed” by you and hasn’t been tampered with.

Thus, once you’ve stored your private key in the TPM, it can natively encrypt files, decrypt files and generate signatures for you – but if the system isn’t in an approved state, or if the request comes from an unfamiliar process, the TPM can simply refuse to cooperate. There’s no need to export the key back out into the operating system where it could potentially be spied upon – in fact, it's possible to lock down the key completely, so that it can’t ever be extracted, under any circumstances.

This means the only real window of opportunity for someone to steal your key is at the point when it’s generated and loaded into the TPM in the first place. Even that tiny loophole is easy to close off, because every TPM has the ability to generate its own internal keys, which can then be used for encryption-based tasks without ever being seen by the OS or firmware. If you subsequently want to pass a computer on to someone else, you can simply send an instruction to the TPM to reset its key storage, and those keys will be gone forever.

This does raise an obvious question: how do you move between computers without losing access to your secure keys? The answer is that, as well as placing restrictions on how data stored in the TPM can be accessed, you can also control whether and how it can be migrated from one TPM to another, with restrictions such as requiring a password or a specific system state.

How Windows uses the TPM

Any application that uses encryption can get a security boost by using a TPM, rather than storing its keys in system memory or on disk. Windows makes this easy with an OS component called the Platform Crypto Provider, which provides TPM-based cryptography services without developers needing to delve into the workings of the module itself. It can also act as a sort of gatekeeper for the TPM: if TPM is instructed only to use certain keys as directed by this process, that prevents rogue processes from performing an end-run around Windows’ security measures.

Probably the best-known popular application of cryptography in Windows is the BitLocker feature, which provides full-disk encryption. To be strictly accurate, BitLocker encrypts your whole disk except for the Windows Boot Manager, which launches as usual when you power on the PC. When it sees that the system disk is encrypted, this little programme then requests the decryption key from the TPM, which recognises the Boot Manager as a trusted process and provides it. This key is used to unlock the Windows startup files – along with all the other data that’s stored on the drive – allowing the computer to boot.

On its own this might sound rather pointless, since the drive is decrypted and booted automatically. However, it means that the only way to access the computer is by starting up Windows – so if you don’t have the right password, you’re locked out. If you try to access the hard disk from a different OS, or using different hardware, all you’ll see is unreadable encrypted data. The TPM can also refuse to provide the key if it detects a change to the boot manager code or to the hardware environment, shutting off the opportunity for malware to infect your startup code.

In general it’s a good thing that BitLocker uses the TPM to lock away its cryptographic key. However, it raises the significant question of what happens if – for example – the host PC dies, and you need to move the hard disk into a new system to keep working. Microsoft has thought of this: when a drive is encrypted, BitLocker generates a 48-character recovery password, which can be used to decrypt an entire drive. For obvious reasons, though, this can’t be safely stored on the drive itself. In an enterprise scenario it will be recorded in Active Directory Domain Services, where it can only be accessed by network administrators, while for individuals it’s stored in your Microsoft account and can be found via the Microsoft website.

On that note, it’s worth clearing up a point of confusion. BitLocker is only available in the Pro and Enterprise editions of Windows 10 and 11, but that doesn’t mean the Home editions can’t enjoy similar protections. Home users can turn on a feature called Device Encryption which does almost exactly the same things as BitLocker: the biggest difference is that it’s much less finicky about changes to the hardware and software environment, and will normally let you boot as long as the TPM is present and correct.

Aside from BitLocker, the TPM also helps power a few behind-the-scenes security features. In the Enterprise and Education editions of Windows, a technology called Device Guard can be used to run applications in a secure virtual environment, from which they can’t access the system software, while Credential Guard stores sensitive data like password hashes and stored credentials in a virtual machine (VM) that’s inaccessible to processes running in the normal OS space.

Elsewhere, Windows’ networking services can use the TPM to confirm that a computer trying to connect to a server is the same one as was originally registered, and check whether it’s using BitLocker and other security features – adding an extra level of reassurance to everyday security as well as remote support and administration.

What’s new in TPM 2.0?

Windows 11 doesn’t just require a TPM – it specifically calls for version 2.0, which should come as standard in most new Windows machines. The difference isn’t enormous, however. Everything we’ve described so far is fully covered by the original 1.2 specification.

What TPM 2.0 does bring is a few technology upgrades. The original TPM specification used the popular RSA encryption algorithm, but the newer version also supports stronger elliptic-curve cryptography. It can also generate 256-bit SHA-2 hashes, which are far harder to crack than the 160-bit SHA-1 ones supported by the original TPM standard. It makes sense that Microsoft would want to standardise on the most secure TPM version available.

There’s another possible reason why Microsoft has chosen to insist on TPM 2.0: it requires a UEFI BIOS, which the company has long recommended as the best choice for Windows. As well as allowing some BIOS functions to be integrated into Windows, UEFI supports the Secure Boot option, which will refuse to load an OS that doesn’t match a recognised hash. This ensures that if any rootkit-type malware tampers with the Windows startup routine, the boot process will fail and you’ll know something is amiss. Secure Boot doesn’t use the TPM, but it ties in with the same desire to improve security at the most fundamental level.

What’s wrong with the TPM?

The most common complaint about Windows 11’s hardware requirements is that they’re unnecessary: for everyday desktop use, the new OS works perfectly well on older hardware. However, there are some who take issue with the very idea of the TPM.

One objection is that it fosters a false sense of security. The TPM concept relies at least partly on the stability and trustworthiness of the OS that interfaces with it, but that’s not something that can be taken for granted. We haven’t yet heard of an exploit that hijacks Windows to compromise access to the TPM, but it’s impossible to guarantee that one won’t ever be found.

It’s also important to realise that TPM techniques can’t defeat all types of attack. For example, they’re powerless to protect against keyloggers which monitor the hardware as you’re typing in a password, or phishing attempts which invite you to enter your credentials into a nefarious website. This may be one reason why Apple has chosen not to join the Trusted Computing Group, and has instead developed its own T2 security chip, which not only handles credentials such as fingerprint and password data but can control the microphone, camera and other sensitive hardware.

Then there are concerns about the potential of TPM technology to prevent users from doing what they want on their own computers. For example, an OS could use the TPM to remotely verify an application with the publisher, and refuse to open it if it doesn’t match an approved signature. Worse, that approval could be rescinded at any point – such as when the publisher opts to stop supporting it, or goes out of business – leaving you unable to use your installed software.

That complaint was common in the early days of trusted computing – the media referred to the TPM as a “DRM chip”, and free software evangelist Richard Stallman derided it as “treacherous computing”. Those issues are rarely brought up today, though – perhaps because it’s become apparent that such restrictions are more likely to be implemented via cloud-based software subscriptions rather than through the TPM.

This isn’t the place to get into that debate, though. What we can say is that, so far, the rise of the TPM has toughened up security without bringing about the Big Brother-esque consequences that some feared. It might be frustrating to see Microsoft cut off so many computers from its latest operating system, but we can’t entirely criticise the company for establishing a new security baseline, and helping to ensure that the coming generation of PCs are as secure as they can be.

If you’ve bought any new computing equipment over the past couple of years then congratulations – it can’t have been easy. From phones and games consoles, to graphics cards and networking gear, the entire industry has been throttled by the ongoing shortage of semiconductors.

Even the biggest companies have been hit – last year, Apple was forced to scale back manufacturing plans for the iPhone, and more than a year after launch it is still virtually impossible for consumers to buy a PlayStation 5.

Is there any sign of things returning to business as normal or should we expect shortages for the foreseeable future?

A big problem for small companies

“If you can't build PlayStations or iPhones, we don't stand a chance,” laughs Sue Ballard, the customer support manager at Offshore Electronics, a contract manufacturing firm in Guernsey. It makes the digital guts for everything from cat-flaps and escalators, to critical hospital equipment, such as power supply units and even ventilators.

“Our efficiencies have gone down the pan,” says Ballard. “We're doing the same job five times over, we've had to change quite a lot of our processes. We're a different company.”

The problem Offshore faces is a severe shortage of components, as one missing piece can cause a knock-on and delay the entire manufacturing process. But the crisis has caused other stresses on the business too.

“Normally you place one purchase order and it's done, the bits come in, you pay it and you don't have to worry about it,” says Dan Attewell, the company’s technical sales director. “But, at the moment, if that stock doesn't come in we've got to go and buy some more. So, you’ve got to raise two purchase orders and you've got to pay two suppliers to get the same amount of bits.”

Even if supplies eventually arrive, it creates chaos for scheduling. “We buy all of the parts in for the month before [the finished item] is due to the customer,” says Attewell. “So we've got a manufacturing window of, say, four weeks. But we might not have all of the bits now until the week before the product's due. Your window gets shorter and your planning is worse, and you then have this massive peak of everything all at once. It's great that we're so busy, but it's an absolute nightmare.”

According to Ballard, it is particularly chaotic because suppliers will often “decommit” at short notice and not deliver what they had promised. It’s a well-known phenomenon in economics known as the ‘bullwhip effect’.

“Information gets distorted and people start to play games with their orders,” says John Fowler, an Arizona State University professor and Motorola professor of supply chain management. “People may order twice as much as they need knowing they're going to get their order cut. And so this whole distortion of information moves all the way back through the chain, which causes the people that are at the very end of the supply chain to get whipped around.”

It’s more than the COVID-19 pandemic

Unsurprisingly, the most significant factor behind the shortages is, of course, the COVID-19 pandemic. “Demand for computers at home increased greatly because people were now working from home and needed to upgrade their equipment,” says Fowler.

It wasn’t just our changing work habits, though, as the policy response of governments around the world had consequences too. “Almost all of the shutdowns came with some form of salary support, some some form of the government giving you money to just take it easy, which meant that unlike any other sort of recession, money was not in short supply,” says Hamza Mudassir, a visiting fellow in strategy at the Judge Business School at Cambridge and co-founder of tech consultancy Platypodes.

A spike in demand was not helped by a shortage on the supply side too, both due to COVID-19 restrictions in countries such as Malaysia, and the need to divert resources elsewhere. “It turns out that the same silicon that is used for semiconductors is also used to make vials for vaccines,” says Fowler. “So that drove up the price for the silicon.”

The pandemic isn’t solely to blame. Another factor is that some of the world’s largest chip suppliers have fallen victim to a series of unfortunate disasters. “There was a plant in Japan owned by a company called Renesas that caught fire,” says Stephane Crosnier, supply chain and operations lead at consultancy firm Accenture. “This plant was supplying about 30% of the worldwide capacity for [some] specific chips.”

Similarly, during 2020, another major fire hit the Unimicron plant in Taiwan, which produces the substrate used to make chips, and a prolonged power outage in Texas shut down major chip foundries around Austin, which are major producers of the wafer on which integrated circuits are printed. “If we hadn't had the pandemic, there was enough shock absorption in the system to handle those incidents, but when it was already on the edge, those things just made it that much worse,” says Fowler.

Slipping back to normality

Given the lack of supply and the enormous demand, the next question is obvious: can’t more chip foundries be built?

“The problem that comes with high-end semiconductors is that they are notoriously difficult to make,” says Mudassir. “The latest Intel processors, the new PlayStation, the new Xbox Series were all designed in a 7nm process that can only be done by a handful of companies in the world.

“The foundries where the semiconductors are built are very, very hard to actually build. It can take billions of dollars and three to four years for you to get to a point where a 7nm or a 4nm foundry is fully functional and running at capacity.”

Mudassir is optimistic that there will be good news – eventually. “Now what we have is a block of time that everybody is stuck in,” he said. “Foundries are being invested in, people are being trained, everybody's working their hardest, but the actual laws of physics are going to prevent you from getting a foundry up and running until there's a major technological breakthrough.”

Consequently, you shouldn’t expect a return to normal buying patterns anytime soon. “When you look at what companies are saying in this space, I think they still expect 2022 to be very, very difficult and are starting to say they're probably not expecting any back to normal before 2023,” says Crosnier, who also warns that further COVID-19 developments, such as a major outbreak in China, could continue to cause problems.

This is in line with what Offshore Electronics are expecting too, though there is clearly still a lot of uncertainty. “I think we're probably still maybe another year away from seeing that really come to fruition,” says Attewell. “We're seeing little glimmers here and there. We're looking for bits every day and [sometimes] someone would decommit us, and the next thing you know the truck turns up here and we've got loads of bits. And even the supplier didn't know they were coming in.”

In fact, fixing the chaotic information environment could be what helps fix the chip supply chains – and make them more resilient. “We are advocating what we call the creation of a supply chain stress test,” says Crosnier.

“You may remember that on the back of the financial crisis in 2008, financial institutions had to undergo financial stress tests to prepare for that, so we are developing something very similar for companies to be able to first get a measurement of their supply chain resilience and, second, identify their point of weaknesses in supply chains.”

In the meantime, there is cause for optimism. Even though it may take some time there is light at the end of the tunnel. “We may very well, in 2023, go from having a shortage of chips to having more capacity than is needed,” says Fowler. “Because when you build a factory, these days, a modern factory can produce a tremendous amount of chips. Intel is building a factory. TSMC is building several new factories, Samsung is building new factories. So you've got a lot of capacity that is going to come online in 2023.”

The chipmaker Broadcom is reportedly in talks to acquire cloud computing company VMware.

Negotiations between the two companies are ongoing and a deal isn’t imminent, according to a Bloomberg report. The deal terms being discussed are unknown for now.

The acquisition would help further diversify Broadcom’s business away from semiconductors and into enterprise software. In the last four years, it acquired CA Technologies for $18.9 billion and Symantec’s security division for $10.7 billion.

Broadcom makes a range of electronics, with its products used in devices like the iPhone or industrial equipment. However, as data centres have become increasingly important, the company increasing its software offering will give it more ways to target this market.

Last year, it held talks to acquire software company SAS Institute, valued at $15 to $20 billion, but the discussions ended without a deal.

The company also aimed to take over rival chipmaker Qualcomm but was forced to walk away from the deal in 2018 after meeting resistance from the Trump administration. One of the concerns around the deal was that Broadcom’s headquarters was based in Singapore. Since then, it has switched to the US and has its headquarters based in San Jose, California, 20 miles away from VMware’s Palo Alto headquarters.

VMware was acquired by storage technology giant EMC in 2004. EMC was then acquired by Dell in 2016 and VMware spun off from the hardware giant last year.

Michael Dell and private equity firm Silver Lake are reportedly the top investors of VMware. Dell holds a 40% stake following the spin-off while Silver Lake, which has previously invested in Broadcom, holds a 10% stake, according to Reuters.

IT Pro has contacted Broadcom and VMware for comment.

VMware completed its spin-off from Dell to become a standalone company once again in November 2021. This helped Dell to raise money to pay off its debts while VMware could look forward to more business flexibility. The hardware giant shed its 81% equity ownership in the company, which created an independent company with a stock market value of around $64 billion.

Microsoft has announced a lineup of adaptable PC accessories which help to augment or even replace a keyboard or mouse for people with disabilities.

There are three different types of devices within the lineup: an adaptive mouse, an adaptive button and an adaptive hub.

Both the mouse and button are designed to support 3-D printed accessories, making them customisable to suit a range of disabilities or ways of working. The core part of the mouse is a small, squarish device that can clip into a palm rest and the end part can also act as a palm rest. There is also a thumb rest that users can switch around so that it can be used either left or right-handed.

The adaptive buttons can have up to eight buttons, all of which can be specifically programmed. In Microsoft's examples, there is a version with two buttons, just to illustrate how bespoke it can be. The mouse can even be used as a joystick, or as an eight-way D-pad. These can be set up as inputs for macros or more complicated keyboard shortcuts, which could be difficult to action for those with motor disabilities. However, as with the mouse, users can also add 3D printer toppers to the buttons to make them fit specific use cases.

The adaptive hub also allows for several buttons, up to four, which can signal instructions to a computer or phone. The hub also has a range of 3.5mm jacks for wired accessories, including hardware made by Logitech and even an Xbox adaptive controller.

These customisable devices will be coming "in the fall" according to Microsoft, though there aren't any details about price, as yet. Still, it is a positive step towards inclusive technology that follows last year's Surface Adaptive Kit launch. That was designed to make existing input devices easier to use.