AMD on Wednesday pulled back part of the curtain on FSR 2.0, the hotly anticipated upgrade of its FidelityFX Super Resolution spatial upscaler/supersampler (that distinction changes depending on the color of the logo).
FSR 2.0 promises to up your framerates for "free," as it were, the same promise made by competing technologies like Nvidia's deep-learning supersampling (DLSS). But which of these framerate-boosting techs will reign supreme? It's Big Green and Big Red, back at it again.
FSR 2.0, revealed as part of the annual Game Developers Conference, accomplishes this through "temporal upscaling," a technique that uses three different vector points in an image (provided to the algorithm by game developers) to upscale a downrendered image's sharpness and overall quality to appear as indistinguishable from native resolution as possible.
FSR 2.0, unlike DLSS 2.3, does not require any machine learning to train its algorithm. However its compatibility is a bit of a mess. What games, APIs, and graphics cards it will work with are all a matter of detail.
We recommend you head over to AMD's full FSR 2.0 blog to get a sense of what will work with what, but in essence: FSR 2.0 will be compatible with a range of GPUs from both AMD and Nvidia (much like FSR 1.0), as well as Xbox consoles, under a specific set of rules and resolutions depending on the underlying graphics chip you're using.
Perhaps the critical blow to FSR 2.0 before it gets off the ground, though, may be that many (including us here at PCMag) were expecting FSR 2.0 to gift its most important component—support for the 1,000+ game-compatible Radeon Super Resolution (RSR)—to players everywhere.
Unfortunately, because of the per-game implementation, FSR 2.0 will be
Read more on pcmag.com