Also did you forget about RE 3 Remake ? XBO Had more Powerful GPU and more ram but it got worst frame rate result i think even worse than PS4 Pro.
They had to lower the resolution(same as PS4 PRO) to maintain framerates
Also did you forget about RE 3 Remake ? XBO Had more Powerful GPU and more ram but it got worst frame rate result i think even worse than PS4 Pro.
They had to lower the resolution(same as PS4 PRO) to maintain framerates
No - nothing to do with IO. Just the PS5 GPU is clocked higher so has a higher fill rate.
From what we’ve heard taking advantage of the consoles IO significantly in games is a fair old way away and it will affect stuff like LODs and pop in not render targets…
Higher clocks are not going to amount to much, Xbox One had higher CPU AND GPU clocks then the PS4 and we saw how that turned out.
Depends on the game I suppose, although by and large modern games are more shader bound than fill rate bound hence why GPU manufacturers focus on shader cores more for their marketing. Although with pixel fill rate your example is not the best example because the clock differences were slight. The better example is despite being clocked lower the Pro had a higher pixel fill rate than the One X due to the Pro having 64 ROPs and the One X having 32 ROPs.
It’s not a magic bullet for the PS5, not only in the console space but PC GPU space we have many examples for higher clock rates for lower powered GPU’s not making up huge differences in power and not even scaling in a linear fashion to improve things.
Yeah, I mean don’t know. Also on overclock thing, ppl thinks the performance gains are linear with the clock rise, there are tons of articles on pc that shows that is not the case, so the 10.28 Tf is in the very best scenario you can think. Maybe I should have checked my expectations, but with my average joe eyes I see this like we have two, one is more powerful then the other, but the weaker performs better or equal. I mean whether it is related to dev kits, or devs forcing parity or any other bottleneck, im not marginally satisfied with what im seeing. But yeah, CoD analysis showed that things may improve.
I think from those I’ve talked to that the difference here is that the PS5 clocks are very high - and in line with the RDNA 2 GPU’s going out. So there is some advantage to that high clock speed and especially early on its easier to utilise that than the broader yet slower CU’s of the Series X. Its not like its suddenly going to magically close a gap - its just that there are instances where the higher speed will be advantageous.
Lets take for example two cards - AMD RX550M and the GTX 1060 Mobile and look at their ACV benchmarks…
At 1080p ultra the AMD card gets 14FPS the Nvidia card - 38FPS.
That’s quite the difference. The AMD is 4.6Tflops the Nvidia card is 4.2. The AMD card has a faster memory bandwidth too - although 2GB less RAM. But the peak fill rate on the Nvidia card is 80 GP/s vs the 53 Gp/s on the AMD card.
If you compare benchmarks for these cards there is an 8% difference but in ACV for example the difference in performance is huge…
It’s not going to by the extent Mark Cerny implied at his GDC talk. However, there will be scenarios where the PS5 outperforms the Series X if a scene happens to be fill rate bound, but this will likely prove to be the exception rather than the norm as time passes I feel.
You should honestly give a better example because you are comparing two completely different architectures. AMD and Nvidia calculate Flops differently for a start. I mean even on AMD side a 5700 trades blows with a Vega 64 despite the Vega 64 having way more raw compute performance.
I think MS played it too safe with clocks after AMD GPUs reveal the series X can easily handle 2.000 MHz imo at least the series S clocks should be higher than 1565 MHz to sqzusee every last drop of that gpu lol
I agree - my point was that you can look at two cards relatively equivalent performance and Tflops (I agree that you can dispute how these are reported) but see huge differences in actual game performance.
yeah but you are using not great architecture(Polaris 12) against one of the best GPU architectures(Pascal)
I understand but a better example would be if you had two cards from the same generation with similar raw compute but one was clocked significantly higher since you are better isolating raw pixel fill rate as the main differentiator.
So, far it seems that the extra 2 TF were a complete waste of time to hype and talk about. For MS sake, let’s hope things improve.
I don’t think there are any actual same generation and architecture examples of this… Which is probably very telling in and of itself.
I’d hazard a guess that it’s a true statement in theory but that in practice it’s cheaper to get more performance out of more compute units then it is to just crank the clocks up on a GPU with less compute units until the benefits of the clock speeds overwhelms the compute unit advantage.
This confusion is really un-necessary.
Seems like the latest patch on series x did some changes for the worse. If I start the Game with the console set to 120hz, sometimes the game somehow does not recognize that and still runs at 60hz. I have to restart the game to fix that. Hadn’t run into that problem before. Also reading comments of crashes after the update which probably are tight to this. Hope they fix that asap, before the game was almost flawless.
The latest patch seems to break something on every platform
So they cant hype their machine?