WYP
News Guru
Last edited:
DLSS2 is literal voodoo, AMD's version needs to match it like for like, And in mainstream AAA games and not cheapo games like Strange Brigade, For me to consider a Radeon purchase.
DLSS2 is literal voodoo, AMD's version needs to match it like for like, And in mainstream AAA games and not cheapo games like Strange Brigade, For me to consider a Radeon purchase.
DLSS2 is literal voodoo, AMD's version needs to match it like for like, And in mainstream AAA games and not cheapo games like Strange Brigade, For me to consider a Radeon purchase.
Voodoo.
![]()
With MS and AMD supporting this that is a big push without a doubt.
On the other hand Nvidia still holds what 80% GPU market share? It's been in use for a few years and has had successful support and applications? It'll be pretty interesting to see how this plays out.
Xbox will undoubtly help make this a thing since its apart of their ecosystem, my thing is what will Sony be using for this? It just further fragments the features of next gen games. Comes down to time/investment/support from Sony and how they interact with Studios.
I've always felt an open platform technology sometimes seems to be better than close platform.
DLSS as great as is it has to be coded for the game.
DirectML seems to be something any game can use.
I'd always go for for a slightly worse open solution to a slightly better proprietary one.
As far as my understanding goes, according to GamersNexus, this isn't supposed to be like AMD's straight answer to DLSS.
At 15:24 in the video below.
Sounds like all AMD is saying here is that "Our version doesn't necessarily work in the same way with library/game support". If they've created a version that works at the driver level or something they wouldn't want it to be associated with the limitations of DLSS.
I think it's also wrong to state DirectML is any more software based than DLSS AngryGoldfish, both are software that is only possible for these use cases because of hardware acceleration for the calculations of neural networks, DirectML still requires hardware support for something like this, in the same way DLSS could be run on a CPU but would become impractical, but hardware support (In terms of additional instructions for accelerating relevant calculations, usually mixed precision fused multiply-add matrix calcs) has been quite commonplace for a year or two now, long before anything useful software wise has been available for it.