Sat. Jun 22nd, 2024



Key Takeaways

  • ARM laptops with NVIDIA graphics can bring gaming-grade performance to the Windows on ARM ecosystem.
  • NVIDIA’s ARM chip for PC gaming handhelds may outperform AMD and Intel alternatives due to technologies like DLSS.
  • NVIDIA’s upcoming ARM SoC with CUDA support may attract professionals seeking AI workloads, challenging Apple Silicon Macs.


With rumors about NVIDIA’s ARM solution, either in cooperation with MediaTek or as a standalone product, getting louder with each new day, it’s only a matter of time before we get an ARM SoC (System-on-a-Chip) featuring NVIDIA graphics. Here’s what I hope we’ll get when NVIDIA chips join the Windows on ARM ecosystem.


ARM-Based Gaming-Grade Laptops Powered by NVIDIA GPUs

The current Windows on ARM laptops, powered by Qualcomm’s ARM solutions, look like respectable options for business-oriented people and those looking for low-power but relatively powerful ultrabooks. However, they aren’t that great when it comes to gaming.

The demos I’ve seen certainly look promising, and I love that Qualcomm seems interested in making PC gaming on its ARM SoCs a thing. Still, I’m not convinced that the first wave of current Qualcomm ARM solutions, the Snapdragon X Elite and Plus, offers good gaming performance.


Sure, the just-released Automatic Super-Resolution (Auto SR) will undoubtedly help with performance, but the numbers shown are far from 60 frames per second. It looks like Qualcomm and Microsoft are targeting 40fps in newer AAA games. That’s decent, but far from impressive.

On the other hand, an NVIDIA ARM solution powered by RTX graphics could allow for some proper ARM-based gaming-grade Windows laptops. Not only would the native resolution performance be miles better than what we’ve seen on Qualcomm-powered Windows on ARM machines, but there’s also DLSS, which is a better upscaling solution than Auto SR. You’re also getting extra frames, thanks to DLSS frame generation.

Considering the great efficiency of the current NVIDIA GPU architecture—the RTX 4050 mobile GPU can use as low as 35W of power—combined with impressive performance at low TDPs—at 35W, the RTX 4050 achieves 43fps at native 1080p in Cyberpunk 2077 with ultra settings—I’m positive that, once we get an ARM chip featuring an NVIDIA GPU, it will blow Qualcomm’s solutions out of the water.


Another advantage NVIDIA has over Qualcomm is the amazing job it has done with graphics drivers over the last 25 years. I’m sure the company will continue doing a commendable job concerning graphics drivers for its future ARM chip.

When NVIDIA joins the Windows on ARM race, we’re definitely getting ARM-powered Windows laptops that can game.

PC Gaming Handhelds Powered By NVIDIA’s ARM Chips

Once we get ARM-based Windows laptops with NVIDIA graphics, what’s stopping handheld gaming PC manufacturers from using that chip in their future gaming handhelds? I reckon the chip will be optimized for lower TDPs, and while the first generation might use too much power to be suitable for the handheld form factor, I’m confident NVIDIA will enter the PC gaming handheld fray sooner or later.

After NVIDIA launches a low-power ARM SoC equipped with Blackwell or Rubin GPU cores, AMD and Intel will have a hard time landing their chips on high-end PC gaming handhelds. Sure, we’ll see handhelds rocking x86 AMD and Intel APUs, but thanks to its dominance when it comes to gaming performance and the advantages DLSS has over FSR and XeSS, most manufacturers will opt for NVIDIA instead of AMD or Intel, given the chance.


Considering that DLSS looks better at lower resolutions (such as 1080p) than XeSS and especially FSR, PC gaming handhelds powered by NVIDIA graphics will most likely offer higher performance and better upscaled image quality than their counterparts featuring AMD and Intel solutions.

We might not see a budget gaming handheld powered by an NVIDIA chip unless we count the upcoming Switch 2 as a budget option, but NVIDIA could end up dominating the high-end PC gaming handheld market.

NVIDIA’s Arm-Based Chip Tackling AI Workloads

Nowadays, Apple Silicon Macs are the best solution for anyone looking to run LLMs locally thanks to their massive amounts of fast unified memory that can pretty effectively play the role of video memory. When it comes to LLMs, the more memory you’ve got and the faster that memory is, the better.


However, no consumer-grade solution can compete with NVIDIA GPUs. They’re the best option for AI workloads at the moment, not only because of their blazing fast AI-related performance—even the low-end RTX 4050 features almost 200 AI TOPS, five times more than Microsoft’s 40 TOPS requirement for Copilot+ AI PCs—and lots of high bandwidth video memory, but also because most AI software is first and foremost built for the NVIDIA CUDA platform.

When NVIDIA debuts its ARM SoC packed with graphics cores and CUDA support, the only limits to AI performance will be the amount and bandwidth of unified memory available to those cores. If both come close to what we’ve got on the Apple Silicon Macbooks and Mac Studio workstations, I reckon many professionals interested in AI will flock to PCs powered by NVIDIA’s ARM chip.

NVIDIA CUDA Debuting on Windows on ARM

The NVIDIA CUDA platform is great for AI workloads, but it’s also the best choice for various other professional workflows, including photo and video editing, GPU rendering, video encoding/decoding, and more. This is why virtually every Windows workstation laptop and desktop includes NVIDIA GPUs.


Windows on ARM laptops powered by Qualcomm Snapdragon SoCs, on the other hand, are pretty good for lightweight workloads and use cases such as content consumption, text editing, web browsing, working in web-based apps, and light photo or video editing. Still, they aren’t that great for serious work.

This is why I don’t think many professionals will switch to Windows on ARM laptops and will instead stick to x86 laptops equipped with NVIDIA GPUs or Apple Silicon Macs. However, once the Windows on ARM initiative gets its first ARM chip packed with NVIDIA GPU cores boasting CUDA support, we might get some proper ARM-based workstation notebooks that will be of interest to professionals.

NVIDIA’s Arm SoC Competing With Qualcomm’s Solutions

Lastly, Qualcomm is currently the only option if you’re interested in an ARM-powered Windows laptop, and being limited to one option is never good news for consumers. This is why NVIDIA entering the Windows on ARM market can only be positive. More competition is always a good thing.



But hey, why only NVIDIA? As far as I’m concerned, the more, the merrier. This is why I hope that AMD, Intel, and even some ARM vendors currently sticking to mobile platforms, such as Samsung, will soon offer their own Windows-compatible ARM SoCs. Well, Intel most likely won’t do that anytime soon, considering the company doesn’t have any ARM chip in the pipeline, but I hope others will follow NVIDIA and make Windows on ARM a thriving market with a number of quality options for all kinds of use cases and budget tiers.



Source link

By John P.

Leave a Reply

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