Older graphics playing cards from AMD and Nvidia could not reap the advantages of AMD’s next-generation upscaling know-how, FidelityFX Super Resolution 2.0 (FSR 2.0). AMD launched a listing of advisable GPUs for numerous resolutions with FSR 2.0, and a few Polaris and Pascal graphics playing cards are notably lacking, together with what continues to be the most popular GPU today, the Nvidia GeForce GTX 1060.
AMD’s new upscaling know-how represents a serious leap for the pink staff. From FSR 1.0 to FSR 2.0, AMD has flipped from a spatial upscaler to a temporal one, and with that reportedly provides an excellent uplift in efficiency and high quality.
The draw back is that it might be too intensive for some older GPUs. With a goal decision of 1080p for FSR 2.0, AMD recommends an RX 6500 XT or RX 590 of its personal playing cards; or a GTX 1070, GTX 16-series or above from Nvidia’s lot.
At 1440p, the suggestions improve to the RX 6600, RX 5600, or an RX Vega GPU or above from AMD; or a GTX 1080, RTX 2060, or RTX 3060 or above from Nvidia.
Then to push a 4K decision, AMD suggests an RX 6700 XT or RX 5700 or above; and an RTX 2070 or 3070 or above from Nvidia.
AMD has briefly spoken about what may decelerate FSR 2.0 whereas in motion, citing FSR 2.0’s comparatively excessive reminiscence bandwidth calls for. It famous throughout a GDC speak {that a} 4K scene can outstrip the Infinity Cache on even an RDNA 2 GPU, which does not bode effectively for older playing cards with out something near an RDNA 2 GPU’s cache or reminiscence capability.
| Goal upscaling decision | AMD GPUs | Nvidia GPUs |
|---|---|---|
| 4K | RX 6700 XT, RX 5700 and above) | RTX 3070, RTX 2070 and above) |
| 1440p | RX 6600, RX 5600, RX Vega and above | RTX 3060, RTX 2060, GTX 1080 and above |
| 1080p | RX 6500 XT, RX 590 and above | GTX 16 collection, GTX 1070 and above |
Although AMD is rolling out an optimisation known as Cache Blocking to enhance cache hit charges. It does so by splitting workloads into multiples to maintain knowledge within the native cache and preserve quicker execution occasions. AMD says L0 cache hit charges elevated 37% on an RX 6800 XT when splitting workloads into three blocks utilizing Cache Blocking.
And that characteristic ought to work throughout GPUs from AMD and Nvidia, not like some others in FSR 2.0 which are RDNA 2-only. There’s possible extra to AMD’s suggestions than reminiscence bandwidth alone, however the omission of the RTX 3050—which delivers decrease reminiscence bandwidth than a GTX 1070—would recommend it is fairly vital for FSR 2.0.
This is not essentially a loss of life sentence for FSR 2.0 on the nonetheless extremely well-liked GTX 1060, nevertheless. It is simply that AMD would not advocate it, as doubtlessly the influence of the algorithm outweighs the advantages it will in any other case provide on higher-end {hardware}. The identical goes for well-liked GPUs comparable to AMD’s personal RX 580 and RX 570.
“Relying in your particular system specs, the system necessities of particular person video games that assist FSR 2.0, and your goal decision, it’s possible you’ll be nonetheless capable of have upscaling expertise on lower-performing or older GPUs than listed under,” AMD says.
The easiest way to learn how these older or extra price range graphics card stacks up with FSR 2.0 might be to strive them out when FSR 2.0 arrives, which ought to occur in Q2 2022, maybe round April.
