Ctrl+Alt+Delete: DLSS still gives Nvidia an advantage over AMD GPUs
Nvidia’s DLSS technology has been one of the standout features for GeForce graphics cards in the past few years.
Using artificial intelligence, DLSS is able to alleviate the workload of your GPU and boost the frame rate of your game. What’s more, Nvidia claims that DLSS is able to maintain a high-quality video quality, so you want need to make major compromises by boosting the frame rate.
This genius AI technology means that Nvidia RTX graphics cards are able to hit higher frames rates than what the GPU hardware is normally capable. This is especially important for gaming in 4K with ray tracing activated, with many systems struggling to achieve a stable performance above 60fps for select AAA games without the use of DLSS.
Of course, FidelityFX Super Resolution launched earlier this year, which is being pitched as AMD’s answer to DLSS. It does a fairly similar job to DLSS, upscaling a lower-resolution images in order to boost a game’s frame rate. However, it has one key difference to DLSS in that it does not use machine-learning algorithms, relying on spatial upscaling instead.
Does this make a big difference? Nvidia believes so. In a recent video, Bryan Catanzaro (VP, Applied Deep Learning Research) explained that while spatial upscaling methods can add additional pixels to a picture, it can’t add more detail.
“The results are often blurry and they often have more temporal artefacts such as flickering and crawling,” he said. “DLSS is temporal method and it uses information from multiple frames to provide increased detail with less flickering.”
Nvidia also showed off some video comparisons to show that DLSS 2.3 is more capable of producing a clear, detailed image, while background objects can look fuzzy or warped with FSR, with some assets even failing to render correctly.
Of course, Nvidia is always going to have a biased view of its own technology, so it wouldn’t be fair to take its word for it.
But if you scour the internet, you’ll see that the majority of experts agree that DLSS provides a superior image quality while also offering similar (if not higher) frame rates.
Deathloop supports both DLSS and FSR, and TechPowerUp has provided a great image comparison which shows how the two upscaling technologies differ. You can see improved detail on the wall with DLSS, while objects in the background look more pronounced.
Honestly, it’s not a massive difference in picture quality, as you really need to look hard at the image to notice the improvements that DLSS offers. But Nvidia’s technology nevertheless seems to be the superior option if you care about having the ultimate picture quality when upscaling.
That’s not to say FSR isn’t worthwhile. It’s still a good alternative to DLSS, especially since it’s available in a huge range of graphics cards. In contrast, Nvidia’s DLSS technology is only available in its RTX GeForce GPUs, which means you’re out of luck if you have a budget graphics card like the GTX 1050 Ti. This is particularly frustrating since DLSS is arguably more useful for low-end cards, since they’re less capable of reaching high frame rates.
But if you’re deliberating whether you should buy a high-end AMD or Nvidia graphics card in the near future, then the performance advantage that DLSS offers is hard to ignore.
Ctrl+Alt+Delete is our weekly computing-focussed opinion column where we delve deeper into the world of computers, laptops, components, peripherals and more. Find it on Trusted Reviews every Saturday afternoon.