Intel and AMD are taking a back seat while Qualcomm is at the center of Microsoft’s AI PC push

To build Microsoft isn’t waiting for Intel and AMD to flex their neural processing units (NPUs) and is pushing its AI PC agenda forward with Qualcomm’s Snapdragon X Elite and X Plus system-on-chips (SoCs).

At a special event ahead of Tuesday’s Build conference, Microsoft revealed that these latest Arm-compatible notebook processors from Qualcomm will power the Windows giant’s next-generation Surface Laptop and Pro tablets. Microsoft’s new Surfaces are, we’re told, just two of 20 “Copilot+ PCs.” Qualcomm says it will launch from June 18 with X Elite and Plus chips.

First teased last fall, the 4nm TSMC-made X processor pair is among the first from Qualcomm to use a Nuvia-derived Arm-compatible CPU core design. If you don’t remember, Qualcomm acquired Nuvia for $1.4 billion in 2021, starting an ongoing lawsuit against Arm along the way.

We’re told the X Plus will have ten CPU cores, and the X Elite will have twelve. The Elite can clock up to 4 to 4.2 GHz for two cores, or up to 3.4 to 3.8 GHz for all cores; the Plus can go up to 3.4 GHz for all cores. If Microsoft is to be believed, Qualy’s chips are more than ready to give Apple’s homegrown silicon a run for its money.

Citing Geekbench 2024, Redmond says its X Elite-equipped Surface Laptop delivers higher multi-threaded performance than Apple’s M3 MacBook Air. It appears that Microsoft has not yet published these results, but similar submissions from Acer and others appear to support these claims.

That said, single core performance appears to be higher on the M3. That would mean that Apple’s eight-core M3 in the MacBook Air is faster per CPU core, although Qualcomm wins by throwing more CPU cores and possibly more power into the benchmark.

It’s not clear how much power the X Elite will need to achieve that goal. When Qualcomm unveiled the X Elite last year, we were told the chip could be configured up to 50 watts. For reference, the M3 is generally considered to have a TDP of around 20 watts.

We reached out to Microsoft for comment on how hot it is to run Qualcomm’s chips, and the IT giant declined to comment on the subject. As usual, we recommend that you take all supplier benchmark claims with a grain of salt.

In any case, it seems that Microsoft’s Surface Laptop, equipped with a 12 core X Elite, 16GB of RAM and 512GB of flash storage, will at least be cheaper than Apple’s latest MacBook Air. So in terms of performance per dollar, Microsoft may have the edge.

In addition to Qualcomm’s new Snapdragon duo, Microsoft is also introducing an emulation layer called Prism, presumably for running older x86 apps on the laptop’s Arm cores. The software giant claims that its latest generation Surfaces are as much as two times faster than the older Arm-based Surface Pro 9 with the help of this emulation layer.

The good news is that you may not need emulation for much longer. Microsoft says native versions of Office, Chrome, Spotify, Zoom, WhatsApp, Blender, the Affinity suite and Davinci Resolve are already ready to run on the machines. Meanwhile, Adobe has committed to bringing its apps to Windows on Arm.

It probably doesn’t hurt that Apple started the move to Arm four years ago, and many of these apps already have Arm64-compatible binaries.

Everything about them TOPS

However, building a performance-oriented Arm notebook SoC is arguably secondary to the NPU at the heart of the X-pair. Capable of achieving 45 TOPS of ML performance on INT8, the .

As it stands, Intel and AMD’s rival parts – the Intel 14th Gen (Meteor Lake) and AMD Ryzen 8040 Series – top out at 10 and 16 TOPS respectively. Qualcomm’s lead won’t last forever. While Intel had previously teased that its upcoming Lunar Lake Core processors would deliver 45 NPU TOPS, it revealed on Monday that the chip’s integrated GPU was capable of pushing an additional 60 TOPS, bringing the chip’s total output to more than 100 TOPS will deliver when it comes onto the market. in Q3.

We expect similar gains from AMD when the next generation of Ryzen processors arrive later this year. Until these chips hit the market, it appears Qualcomm has the AI ​​PC market cornered.

Compared to Apple’s M3, Qualcomm claims its silicon has up to 2.6x higher NPU performance. While the Neural Engine in the M3 maxes out at 18 TOPS, Apple’s newly announced M4 SoCs are capable of pushing this to 38 or perhaps even higher. Unfortunately, that chip is only available in the expensive iGiant iPad Pros.

As for what you’ll do with all those NPU TOPS, Microsoft says they’ll enable a variety of new AI features for Windows. This includes Cocreator for generating on-device art in Paint; and live captions for things like video calls. And then there is Memory…

What’s Going On With This Windows Recall?

Microsoft decided to use this pre-Build event to show off Windows 11’s Recall for Qualcomm-powered Copilot+ PCs and any other computer that meets Redmond’s hardware requirements. The feature – demonstrated here – is said to be currently in preview and awaiting user feedback.

Recall is expected to automatically record everything you do on your desktop – from web pages visited to app activity and conference calls – and store it locally encrypted so you can search through it and retrieve it again using AI. Right now it seems to mainly take screenshots of your desktop every few seconds and save them for future searching. If you vaguely remember doing something on your PC recently, you can query it using text, or drag a slider on the timeline and recall it.

It screams of a privacy nightmare, although Redmond insists it all stays private on your PC. From the frequently asked questions:

We’re told that users can use Recall’s settings to ‘make choices about which snapshots Recall collects and stores on your device. You can limit which snapshots Recall collects; for example, you can select specific apps or websites to visit in a supported browser. to filter out your snapshots.

“Additionally, you can pause snapshots on demand via the Recall icon in the system tray, clear some or all snapshots that have been saved, or delete all snapshots from your device.”

Microsoft says Recall won’t snoop on private browsing in Edge, won’t pick up DRM-protected content, but will likely suck up things like passwords, financial information and the like as you use your PC unless you take action like the above to stop it.

According to Microsoft, you need at least 256 GB of storage space, of which at least 50 GB is free; in such a configuration, Recall’s default usage will be 25 GB, which can store approximately three months’ worth of snapshots. This can be adjusted and old snapshots can be discarded if necessary. It’s a bit like the Windows 10 timeline that was scrapped in 2021.

Microsoft has also touted several apps that have added support for NPU acceleration, including Davinci Resolve, Cephable, and CapCut.

Against the flow

Qualcomm’s X Plus and X Elite promise impressive performance and the NPU grunt Microsoft needs to advance its AI PC agenda, but the chips are also at odds with Arm’s trajectory.

As we mentioned earlier, Arm isn’t too happy that Qualcomm acquired Nuvia and is now designing its own Arm-compatible CPU cores; both Qualcomm and Nuvia are Arm licensees, and Arm is not happy with the way Qualcomm has adopted Nuvia’s Arm-compatible technology. So unfortunate, in fact, that Arm sued Qualcomm for the whole debacle, arguing that the Snapdragon maker had violated its architectural license with Arm and had to negotiate a new one to use Nuvia’s designs, no doubt at a higher price.

Remember, Arm has asked the court to order Qualcomm to destroy its Nuvia-derived designs, which would throw a pretty big spanner in the works, not only for the Snapdragon house, but also for its partners like Microsoft, if the legal battle would break out. That.

Nuvia’s cores are also a step backwards as far as some in the Arm world are concerned. As we understand it, the Nuvia-derived CPU cores at the heart of Qualcomm’s X-series SoCs are based on the older Armv8 architecture and rely on a combination of GPU and NPU acceleration to accelerate AI workloads , along with Arm’s usual NEON CPU-level instructions.

In contrast, Arm has been building AI-accelerating features – such as Scalable Matrix Extensions 2 or SME2 – into its newer Armv9 CPU architecture in recent years. These are already used by Apple in its M4 Armv9 SoCs.

Qualcomm’s decision to move away from Armv9 – the tech giant has previously used the architecture in its smartphone SoCs – and the CPU-level improvements in favor of offloading AI work to NPUs means the Arm world is left with a potential gap is faced: on the one hand, Microsoft and Qualcomm today developed Armv8 processors with custom units to deal specifically with ML code; and on the other hand you’re trying to get people onto Armv9, where AI accelerating instructions are baked into the CPU architecture by default.

Considering that Microsoft has thrown its support behind Qualcomm’s Armv8 chips in its quest to push AI functions out of the data center and into the customer; Apple is Apple; Google in love with its own on-device ML accelerators; and Samsung and MediaTek float in the middle, and with Arm and Qualcomm at legal odds, we sympathize with developers – from OS and framework makers to app builders – trying to achieve maximum portability and maximum performance on Arm. ®

Leave a Reply

Your email address will not be published. Required fields are marked *