IIRC the article referenced a quote from someone at AMD when asked about an answer to DLSS. Still looking for it though.
Also no worries, I don’t mean to insinuate you’re calling anyone a liar, I just think there is misunderstanding somewhere. To me, widespread misunderstanding over the tech behind FSR doesn’t have anything to do with the widespread understanding that it’s a competing feature, because it is. What I mean by that is both features are intended to provide better IQ, even if different methods are used.
Remember that demo showing Spiderman loading? Later in one of the articles for Wired Cerny confirmed that the final ssd was faster than the one used for that demo.
IIRC the demo was shown around the second half of 2019 and the interview was around March 2020.
BC games load and decompress their data from storage using CPU cycles. Most of the improved IO stack in the PS5 is idle in this case and unused. CPU may also throttle in some BC mode
Which DirectX API loads data from storage? These functions all assume everything is already in main memory.
The potential is not very good though. And lets not pretend that people have believed that FSR is like DLSS - they have. And now we can see its not and is merely an upscaling technique that probably won’t be used very often given the results thus far and limited scope to improve it.
Receiving final hardware that matches target spec in terms of speeds and IO design is not the same thing as changing mid-way through development. Prototype system hardware rarely runs at full spec in hardware or software for any platform.
I’m not sure, I’m not a game developer. There must be something about the way data management is handled on the OS or API side for there to be an automatic change without any developer involvement. The CPU explanation doesn’t make sense because BC on the PS5 does use the increased performance of the CPU/GPU. We can easily see this by the huge boost in performance games enjoy when played on the PS5. Not to say the increase in CPU speed doesn’t greatly increase loading speeds in general, there just must be something on the Xbox or Playstation side that prevents the same boost. Maybe the BC method used utilize the CPU differently and that’s where we see the gap?
Sure we can. The cross gen tools for PS4 games basically removes the 30fps cap. So the PS4 Pro games, like Days Gone or GoT, that now run at 60fps. That’s a really nice boost.
Yeah but they are still recognized as PS4 games. So in a way, similar to Gen9 aware games on the Xbox where cross gen tools are used to extract some extra features when running on a new gen console. In the end, these games are still run on last gen APIs and recognized as last gen games, only slightly modified.
Yes but that modification, while not vast, is what allows for the data transfer to be much faster, no? I might not be understanding the discussion entirely here though as I just woke up. >.>
Possibly. With those games still being recognized as PS4 titles, there’s a good chance it would be under the same CPU/GPU limitations regardless, only now without the same 30fps cap.
Edit: A good thing to look at with these titles specifically is to see if they enjoy larger improvements to their load times over what other BC games receive. I honestly never paid that close enough attention to them yet.
Edit #2: Looking at the load times differences for Days Gone, there is about a 6 second difference (19 seoncs vs 25) so it fits in line with loading improvements we see in other more standard PS4 games running in BC. This suggests these PS5 enhanced games don’t utilize the CPU more so than other PS4/Pro games running on the system.
Because they are open world games. Those are almost always CPU bound on last gen systems. You’re not getting an open world game, with 300+ npcs on screen at once, running at 60fps on those jaguar CPUs.
