We may earn a commission when you buy through links in our articles. Learn more.

Arm just brought Nvidia DLSS-style upscaling to mobile devices

Using AMD FSR 2 as a starting point, Arm engineers have developed an upscaler that should help boost performance or save battery life.

arm asr

Ever since Nvidia first unveiled its DLSS 2 upscaling technology and suddenly brought a viable free way for gamers to increase the frame rate of their games, it has brought about a step change in the way gamers and the industry thinks about gaming performance. To that end, Arm has just announced that it has developed its own DLSS 2-like upscaling algorithm, which will work on GPUs for mobile devices, calling it Arm Accuracy Super Resolution or ASR.

Arm, which designs the baseline architecture of the mobile chips used by the likes of Apple, Qualcomm, and Samsung, isn’t a company we’d normally associate with the best graphics card innovations for desktop gamers. However, its architecture sits at the heart of the new and much-hyped Qualcomm Elite X laptop chips, which are establishing themselves as the first truly viable option for Arm-based Windows laptops.

arm asr nvidia bistro scene

The new Arm ASR tech, then, may ultimately be inspired by Nvidia’s original DLSS upscaling innovation, but is actually based directly on AMD’s competing FSR 2 tech. With AMD having made FSR open source, Arm has been able to take that starting point and tweak it to work on its own Immortalis-G720 GPU.

In its blog about the new release, Arm explains how there are already existing upscalers that can work on Arm GPUs, such as Qualcomm’s Game Super Resolution (GSR), but that these are based on spatial upscaling, rather than temporal upscaling. That’s where the upscaling algorithm only works on each complete frame in isolation, using techniques such as edge detection to inform the upscaling.

Instead, FSR 2 and now ASR use what’s known as temporal upscaling, which is where the algorithm compares the just-rendered frame with the one that came before and uses it to help inform the upscaler. The game engine also passes on extra information about how the player and objects in the scene have moved, which further helps inform the upscaler. All this extra information means that temporal upscalers offer vastly better image quality than spatial upscalers.

In developing ASR, Arm notes that FSR 2 “produces fine results but with an implementation that is only affordable on PC and high-end consoles,” though it doesn’t go into detail about what changes it made to make ASR more suited to mobile GPUs.

arm asr performance

To demonstrate the new upscaler, Arm used the openly available Nvidia Bistro scene (shown at the top of the page) to provide a performance comparison between running the game natively and using FSR 1 (a spatial upscaler), GSR, FSR, and ASR.

In its performance charts, Arm shows how ASR can deliver a frame rate increase of between 38% and 53%, depending on the upscaling ratio (the difference between the initial render resolution and the output resolution) compared to running at the native resolution. Meanwhile, the other techniques all result in a lower frame rate for any given upscaling ratio.

arm asr power consumption

Arm also shows how ASR is much more efficient than FSR 2, using as much as 73% less GPU load for any given upscaling ratio. In turn, this results in significant power savings.

Arm also demonstrated the image quality difference between ASR, FSR 1, and GSR, though it notably didn’t include FSR 2 in this comparison. Nonetheless, these images clearly show the advantage of using a temporal upscaler like ASR over spatial upscalers.

arm asr image quality

So what does all this mean for PC gamers? Well, nothing for desktop PCs at the moment. There are no desktop gaming systems that use Arm GPUs. However, for those Qualcomm Elite X laptops and future Arm-based CPUs/GPUs, this could be a great addition.

In the meantime, you’ll have to rely on classic AMD FSR and Nvidia DLSS if you want the best upscaling options on the best gaming laptops right now.