HUB recently did some testing of DLSS vs FSR, many of you will have seen. the best FSR could manage over 26 games was a tie, never once besting DLSS outright.
Off the back of that, HUB opted to test DLSS vs Native, and since FSR never won a comparison vs DLSS it wasn't worth coming in a 3 way shootout, just DLSS vs native.
Here's the table of results below. Tim going on to say that yes indeed, DLSS can give better than native results, broadly on par results, and of course lesser looking results, however on a balance of IQ and performance gained, DLSS is
clearly the desirable one to enable if it's better than native and anywhere between much better and tied, and in his opinion still often desirable over native when it loses, if you aren't CPU limited or otherwise not gaining any performance from DLSS. Of course, if Native had better TAA, it would look better, hardly a shocker there, but Dev's virtually never go back and update TAA - probably the best version of that argument is to use a mod to get DLAA (or native DLAA support), or FSR and ratchet up the internal res to 0.9x-1x to just use it as AA. Tim also rightfully adds how easy it is to drop a new DLL in, which would have changed
at least one result (RDR2), if not more of these results to favour DLSS, or tie. Tim also adds that developers should be updating games to have the latest/best DLL, and same for FSR versions - especially when the game is still actively being patched and updated.
Interesting results and it's great to see what I've been saying for quite some time now be thoroughly corroborated, which is; 'DLSS
can look as good or better than native'. Naturally, I game at 4k where this tech is at it's most useful and thoroughly improves a given cards capability to drive that res.
View attachment 291888