DonFerrari said:
Can't disagree with your explanation. Just wanted to point out that being both architectures balanced (which is what I expect them to be) then texture quality would be PS5 advantage due to the capacity to stream larger (better) textures and the geometry is something uncertain since as you explained XSX have likely more capacity to draw geometries but then Nanite like feature is dependant on the SSD and I/O speed.
|
Playstation 5 could have better geometry capabilities. It depends on how many geometry units each console has (I am just waiting on Big Navi to know more details) and the clockspeeds they run at.
Every component on a GPU runs at the core clock or a multiple/division of that... The Playstation 5 undoubtedly has an advantage in clockrate... So whatever units are "Kept equal" the Playstation 5 will have an advantage in.
For example... The Playstation 5's GPU has a higher Pixel fillrate than the Xbox Series X as they both have the same number of ROP's, it's just the Playstation 5's ROP's operate at a higher clock.
DonFerrari said:
The great question that I believe you agree depends on the games revealed along the gen is how much the I/O interface can help the result. As said if the SSD can send better quality assets due to the speed advantage then perhaps the GPU and CPU can do a little less work could it alleviate some of the disparity? I do understand SSD doesn't do any computation so the performance GAP will always exist, but could the optimization done by devs take care of some of it? Probably not and for the whole gen we will see some advantage on the pixel count, framerate consistency, but for games where both can achieve 4k30fps where would the 10-20% difference go? What type of effects could be leveraged on this type of GPU difference in PC?
|
There is a ton that can be done with that compute disparity, better lighting, shadowing, particle effects, physics, procedural generation, post-process, anti-aliasing, anisotropic filtering and so forth.
It's not an insignificant disparity in compute, just like the SSD isn't an insignificant disparity.
The SSD isn't going to reduce the GPU's work though... That isn't how data or processing works.
The SSD is assisting in the delivering of the "work" to the GPU and does so at a faster pace than other prior methods... One of the largest investments a GPU or CPU manufacturer does isn't actually making the CPU/GPU faster, it's ensuring that the data is available for processing. It's making them more efficient by ensuring there is always work available.
Extremely fast CPU's for example will actually spend the majority of it's transistors on technology to keep the data close to the cores with caches, or to guess the data required for processing ahead of time via things branch tree prediction and so forth...
Because retrieving memory from RAM is a ton of wasted clock cycles.. We could be talking 50~ billion or more potential instructions going to waste here on a modern design, hardly ideal.
... And that is retrieving data from RAM. - The hit from retrieving data from disk is an even more pronounced issue... But thankfully doesn't happen very often.
In short, the SSD will ensure that the RAM is consistently fed with the required data and will be more effective at doing so over time than the Xbox Series X.
But if we were to put it in the Unreal Engine 5 demonstration terms... The Xbox Series X will hold significant advantages in the lighting/Lumen department, where the Playstation 5 should hold the advantage with Nanite... That's not to say the Xbox Series X won't still have an impressive showcase with Nanite or the Playstation 5 can't impress with Lumen, but that is generally where the strengths of each console lay.
For the most part though in cross-platform titles, the Xbox Series X may take advantage of higher PC graphics settings and sink it's extra performance chasing higher framerates and resolution and the Playstation 5 will have shorter load times when the CPU or other component isn't a bottleneck. (Loading is more than just storage remember!)
It's the first party AAA where the strengths/weaknesses of each hardware platform will become more pronounced.
DonFerrari said:
Great, because usually on VGC (I don't have a gaming PC) we see comparisons of the fps on normal, high, ultra, etc. But I was curious on a 10-20% difference on horsepower (here let's pretend the TFlop is a direct correlation to general power) and both games have to achieve "the same" framerate and resolution how much could be done better with the power discrepancy, because we know that devs try to minimize their cost and time so it needs to be good bang for buck investment of the power dif.
|
Need to account for the differences over time, this is stuff that black and white spec sheets can't really properly explain.
In a deferred renderer for example various operations get deferred over time... It's why Battlefield 3 looked as good as it did on the Xbox 360/Playstation 3. (Plus lots of baked information too.)
So that 2 Teraflop/20% disparity will grow over multiple cycles to potentially 6 Teraflops, just like the SSD GB/s bandwidth disparity grows over multiple cycles.
But then I used the key word "baked".
The Playstation 5 can bake GI/Shadowing into some extremely high resolution assets that will look the goods as it has the bandwidth available to pull it off.
