Digital Foundry Article Technical Discussion [2021]
- Thread starter BRiT
- Start date
- Status
Which is part of the reason I don’t mind being about 6 years behind in my games backlog. By the time I get to a game the definitive edition HD texture raytraced version is likely available with (hopefully) most bugs squashed.
How can you make such categorical conclusions about the architecture of consoles when we have not yet seen a single game of the next generation.
All games released at the moment have engines developed for architectures of a decade ago, which rely heavily on ROPs and CPU/GPU frequency. With a full transition to the "next generation", everything will change a lot.
The hardware accelerators Mesh shaders, SFS, ML have not even been used yet (not to mention that they will still have to be studied properly).
RT, VRS and the capabilities of the new CPU were only superficially involved.
See what difference in graphics quality, performance, memory consumption and data loading speed can be achieved when these technologies are used:
Quite a bit can be extrapolated from PC RayTracing performances and knowing the pitfalls.
snc
Veteran
is this post from 2010 ? :d most games even crossgen are heavy on computing, some like metro exodus 90% compute 10% raster
Which is part of the reason I said at and after launch.
But the fact is, it still has months to go and optimisation generally happens at the tail end of development. So all versions can change, we don't know where each version is in development especially when you consider lead platforms.
All you can do is say at launch this is how they perform which is totally fair, regardless of state of development between platforms.
davis.anthony
Veteran
Why are you talking about pixels or throughput?
I simply made a statement that at times PS5 can have a frame rate that's 44% faster then XSX's..... which is true according to the frame rate numbers on the DF video.
Hitman is an example of a less stable framerate, but the reason those scenes are ROP limited is because of the higher pixel output. I grabbed that 45fps number because it was the one I saw most often on the lower end in the scenes in the grass. Lots of times it's in the 50s, and can go as low as 41. It wasn't meant to show some advantage for Xbox, just to point out that if you actually calculate the rendered output, the 2 machines are close to each other. Maybe 8% higher here, maybe 2% lower there. But definitely not 44% more or less rendered output. Valhalla is a better example I think because IIRC launch pixel counts were similar and it's framerate issues were throughout the entire game and not limited to a single scene. And it's solution was to lower the resolution on the platform with problems.
And I'm not saying it isn't true that the framerate is 44% faster. Your statement wasn't that simple. You also said...
That's only a 44% increase in frames but doesn't consider what's in the frames. I can do 44% more curls than The Rock. Just don't ask what weights we are each using.
Why am I talking about pixels or throughput? Because this is a technical forum discussing the ROP performance of 2 pieces of hardware. Pixels and throughput are a big part of what that's all about.
davis.anthony
Veteran
So with that said it would be appreciated if you factor in the 60fps cap restricting PS5's results when comparing pixel output in Hitman.
There’s no value talking about PS5’s hitman performance, as it’s doing what the developers set out to do, which is hold a steady 60fps.
Saying that there is a 44% frame rate advantage greater than any paper advantage is pretty console warrior territory. There is no “frame rate” advantage that PS5 holds over XSX in the argument, we use PS5 to showcase an obvious issue with XSX bottleneck for that moment. Gain insights into how developers may have rendered certain things. Meaning that the XSX is not performing as the developers desired for such a heavy penalty, losing nearly 50%. Which is not the same as just hitting the limits of the GPU and seeing some minor frame loss.
There is an obvious bottleneck they hit and we can use that information to gain insight into XSX hardware performance, we do not gain insight into PS5 hardware performance. PS5 does not hold a 44% frame rate advantage here because XSX is clearly sitting idle without it’s hardware being utilized.
davis.anthony
Veteran
Good job I've never anything of that ilk then.
DF Article @ https://www.eurogamer.net/articles/...-definitive-vice-city-san-andreas-tech-review
GTA Definitive Edition: Vice City and San Andreas add to the disappointment
And the new patches do not address fundamental problems.
We've already looked at the opening entry in the Grand Theft Auto Trilogy Definitive Edition - and found the remastering to be anything other than definitive. Shoe-horning the RenderWare-based originals into Unreal Engine 4, presumably in order to tap into the modern technology's advanced rendering capabilities, delivers an experience that at its very best falls short. The mixture of old assets, AI upscaled textures, geometry-enhanced/smoothed vehicles, remodelled characters and modern rendering sensibilities is jarring to say the least. GTA3 was the oldest game and potentially showed off the biggest possible boost but what about Vice City and San Andreas? And what about the recently released patches - is the situation made any better?
Before we go into depth on the individual games, it's important to put the patches into context. Since they arrive relatively soon after the release of the trilogy, we should expect them to have been in development for some time - likely before launch, before the backlash. The patch notes suggest as much, ticking off fixes for a range of bugs as opposed to addressing the fundamental issues raised when the game launched. Our work on the games was mostly carried out on the unpatched launch code, but we've spent significant time re-examining the post-patch Vice City and San Andreas. Unfortunately, none of our issues with the game had been addressed and performance was also unchanged. Interestingly, there is a mention of improvements to the rain effect in San Andreas in the patch notes but even this seems to run in much the same way as before.
...
GTA Definitive Edition: Vice City and San Andreas add to the disappointment
And the new patches do not address fundamental problems.
We've already looked at the opening entry in the Grand Theft Auto Trilogy Definitive Edition - and found the remastering to be anything other than definitive. Shoe-horning the RenderWare-based originals into Unreal Engine 4, presumably in order to tap into the modern technology's advanced rendering capabilities, delivers an experience that at its very best falls short. The mixture of old assets, AI upscaled textures, geometry-enhanced/smoothed vehicles, remodelled characters and modern rendering sensibilities is jarring to say the least. GTA3 was the oldest game and potentially showed off the biggest possible boost but what about Vice City and San Andreas? And what about the recently released patches - is the situation made any better?
Before we go into depth on the individual games, it's important to put the patches into context. Since they arrive relatively soon after the release of the trilogy, we should expect them to have been in development for some time - likely before launch, before the backlash. The patch notes suggest as much, ticking off fixes for a range of bugs as opposed to addressing the fundamental issues raised when the game launched. Our work on the games was mostly carried out on the unpatched launch code, but we've spent significant time re-examining the post-patch Vice City and San Andreas. Unfortunately, none of our issues with the game had been addressed and performance was also unchanged. Interestingly, there is a mention of improvements to the rain effect in San Andreas in the patch notes but even this seems to run in much the same way as before.
...
davis.anthony
Veteran
I don't think they needed to do as much as they did, the biggest issue with these older games was frame rate and draw distance.
Lock to 60fps, completely eradicate pop-in and they would have been half way there.
Lock to 60fps, completely eradicate pop-in and they would have been half way there.
Completely agree. I think (seriously) that GTA3, VC and SA look better on the PS2, in special when played on a CRT or/and when forcing higher output resolutions which where huge drawbacks.
How could they put out a trilogy to the one of, if not the most popular and best selling games ever in such a bad 'remake' shape, looking and playing worse than it did 20 years ago on 1999 hardware on todays premium consoles. See what bluepoint/sony did with demon souls..... Or what community modders do to remakes/remasters of GTA games.
The whole trilogy just looks out of place, the rain effect.... everything basically. Aside from performance, at times, worse than my 21 year old PS2 does. GTA3 for example, it looks more polished and better graphically before the remake on 6th gen consoles, even character rendering.
Lets just assume the trilogy was a joke. And indeed, they could have locked the fps to a smooth 60, upped the draw distance somewhat and perhaps added higher texture quality, improved assets etc like they did to GTA3 and VC for the OG 2001 Xbox 18 years ago. Yes, 3 and VC both did get a locked 30fps, higher draw distance, textures, lighting and assets improvements, on the OG xbox launched in 2001.
davis.anthony
Veteran
Studio's are getting over excited with remakes/remasters these days.
Alan Wake Remake made the character models look horrendous.
The Crysis remaster completely changed the TOD to the point where the original looks better (imo)
There'a line in the original Jurassic Park that I always think about when it comes to remasters....
"They were so busy wondering if they could that they never stopped to think if they should"
''spared no expense''
davis.anthony
Veteran
- Status
Similar threads
