R* added a load of traffic, if they had left that the same I've little doubt we'de be seeing 60fps - certainly the ability to lock a solid 30FPS 100%
-
Hey Guest. Check out your NeoGAF Wrapped 2025 results here!
Digital Foundry: GTA V PS4 and Xbox One compared in new frame-rate stress test.
- Thread starter Hanmik
- Start date
R* added a load of traffic, if they had left that the same I've little doubt we'de be seeing 60fps - certainly the ability to lock a solid 30FPS 100%
In order vs out of order should not really matter when porting if your in order code is already pretty good.
The biggest difference is that Power Architecture is RISC (reduced instruction set computing) based but X86 is CISC (complex instruction set computing) based. These architecture work in a fundamentally different way so it requires a lot of work to port it over.
Ashler
Member
This needs to be addressed (if at all possible, CPU overclock via firmare?) by Sony.
If all these new games will be bottlenecked by the CPU, then it doesn't matter that much if their GPU is 40% better (well, it matters as they can render everything with more detail).
I don't know the thermal threshold they have due to their smaller box, though. It might cause them more issues (heating, etc) than solutions.
For a console that is supposed to be the clear superior performer, its might start to be a thorn on their side if we start seeing more CPU intensive games.
If all these new games will be bottlenecked by the CPU, then it doesn't matter that much if their GPU is 40% better (well, it matters as they can render everything with more detail).
I don't know the thermal threshold they have due to their smaller box, though. It might cause them more issues (heating, etc) than solutions.
For a console that is supposed to be the clear superior performer, its might start to be a thorn on their side if we start seeing more CPU intensive games.
JonathanPower
Member
Honestly, to run such a basic OS you probably need just a fraction of the power of a single Jaguar core. And surely you don't really need more than 1GB of RAM.
It is not a CPU bottleneck as you can see if you read my larger post above.
Ashler
Member
If we count CPU only it is. From your post:
Xbox 1 = 1750 * 6 * 8 = 84 GFlops.
PS4 = 1600 * 6 * 8 = 76.8 GFlops.
IF this is a purely CPU bottleneck, it won't matter if their GPU or memory architecture is better or if the raw crunching numbers are better. The CPU will always 'chug' (its not that wide of a difference, hence the '') along for the ride.
I'm not an actual developer so I could be wrong. So in my uneducated opinion, there are already two separate instances (AC U and GTA V) where the supposed inferior machine is out performing the other. Both cases have been linked to the CPU difference.
Exactly. It's not difficult to understand. GTA V is game that was written bottom up for different architecture. We have individual benchmarks for each component, which provides a general idea of what to expect, but getting each core to work in parallel and sing in harmony is a serious challenge. A Sony first-party studio described how, in a single situation on PS4, they literally wrote ten different algorithms before deciding which worked best. So it's entirely expected that many other development teams will be making small improvements also.
Anyone can dance (badly) but getting individual cores to perform in rhythm like this takes time and investment. That applies to both systems.
I meant this post.
If the CPU was a bottleneck for 100% of the frames shown in a game then the best the Xbox One could do vs the PS4 is a 9.4% advantage in framerate assuming it is below the FPS cap. This could still lead to the PS4 having better graphics though since it does have the GPU advantage. The fact there are instances where the FPS is 20% + higher suggests that there is something else going on.
That is not how it works. Also, the CPU-difference is minimal.
Mr. X-Men?
On topic, good to see the PS4 version appears to be hitting that 30fps nearly 100% of the time. I have noticed those drops driving through down town during the day, would be nice if they could maybe patch in a fix for that, although it isn't really too bad when compared to something like RDR on the PS3 and it appears to be quite random.
As for the article, well as I've seen from comments here and on eurogamer, it makes it out as if the PS4 is the place to play for shooting and the Xbox One is the place to play for driving, when the videos show that the PS4 version consistently matches or out performs the other platform everywhere apart from those random daytime intersections. Poorly written, gotta keep up the narrative of the "bottlenecked" CPU even if there is no hard evidence or data to support that claim at the moment (not that the CPUs on both machines could have been a bit beefier).
All in all it seems to be a bloody good port on both platforms, here's hoping they do a good job for the PC too.
funkystudent
Member
The article really doesnt seem to match the performance shown in the video. XB1 seems a heck of a lot more unstable in it then the write up makes it seem.
Also where is the Far Cry / Dragon Age comparisons from DF. Instead they have spend a week drop feeding GTA over and over. Really not liking the way DF have been handling themselves recently.
Also where is the Far Cry / Dragon Age comparisons from DF. Instead they have spend a week drop feeding GTA over and over. Really not liking the way DF have been handling themselves recently.
Deaf Spacker
Memb
This thread is amazing already.
It's a stress test designed to test the hardware to the limit, in a normal gaming session I cannot see that many dips in the framerate.
It's a stress test designed to test the hardware to the limit, in a normal gaming session I cannot see that many dips in the framerate.
Problem with this is you can't just add these numbers. These numbers are dependant in a special meaning on each other and especially raw numbers don't have proportional relation to performance (in this case, gaming performance).
Likewise you can't single out random events in games - compare FPS and use those small examples to make an issue of something that doesn't exist (when you compare the rest of the data which shows expected results).
Clearly it's false to claim the XBO performs better in certain circumstances because those circumstances are impossible to re-create 100% accurately.
alexandros
Banned
I don't know man, the screenshot on page 10 seems pretty clear. That is a 5 frame difference in a 30 fps game. It's substantial.
Hence why I qualified it by saying it was raw number crunching performance. Only time you will see that sort of performance is if you were to run some sort of folding application or mining application on the consoles.
In reality they will be lower than their theoretical peaks due to other bottlenecks in the systems but assuming fairly even utilisation the performance gap will be in this ballpark.
It is fine to talk about relative differences like a 9.4% CPU advantage and a 40% GPU advantage but when you have posters or even reputable websites talking about CPU bottlenecking it is important to put the relative differences into context with total system performance.
Michael F. Assbender
Banned
Wow, that had almost nothing to do with what he said.
Thing is, there are so many variables involved in this context we don't know either, meaning: we don't know if it's a bottleneck nor if it's not.
Without proficient data (profiling data) we don't know why certain situations are like in the article. And then we even don't know how the code looks like, we could only guess given on the data.
Mmm yeah a single event captured by a screenshot it's an undeniable fact how performance work. Right.
Messofanego
Banned
People wait to get a current gen port of a last gen game, still get sub-30fps in quite a few spots lol.
Looks like ppl waiting for PC version win out after all
Looks like ppl waiting for PC version win out after all
People wait to get a current gen port of a last gen game, still get sub-30fps in quite a few spots lol.
Looks like ppl waiting for PC version win out after all
Sssh. Hush your mouth with that superiority complex.
alexandros
Banned
I think it did but I might have misread.
People wait to get a current gen port of a last gen game, still get sub-30fps in quite a few spots lol.
Looks like ppl waiting for PC version win out after all
Do you know GTAIV on PC? Even with tenfold the power of last gen consoles, I can not run it at stable 60FPS. Even 30FPS still dip from time to time. It's just how Rockstar Games work. They probably had it running at stable 30FPS and just said "well fuck it, it wouldn't be our game if it wouldn't dip, people expect that" and then added some shit code.
Well with the Athlon 5150 vs the 5350 you can see that a 28.1% CPU advantage does not lead to any above error margin FPS gains in any game, even open world titles like Sleeping Dogs. You can also see that in a specifically designed CPU limited test such as the Unity Draw Call test this 28.1% CPU advantage only results in an 11.7% higher score.
That leads me to believe that there is very very little scope for the PS4 CPU to be bottlenecked and the Xbox One CPU not to be, and if both are bottlenecked you would expect an at best 9.4% advantage not the 20% we have seen in certain scenarios.
You would also expect that to show each and every time you went through that specific place in the game but at the 3:43 mark in the video there is that scene where both consoles are holding the same 25 FPS rate in a busy intersection with lots of traffic. The only part of that video where the Xbox One outperformed the PS4 was towards the start where they were driving through the intersections and the traffic scenes were different on each console so there is nothing conclusive with which to base this CPU bottlenecking theory on.
Messofanego
Banned
Aren't explosions esp in Trevor missions and high speed chases likely to be present in a normal gaming session? That's kind of what people who play GTA tend to get into, mayhem.
Windom Earle
Banned
I wonder if Rockstar will ever make a console sandbox game with a perfect framerate. Even the mighty RDR on 360 had issues in some of the towns.
GTA6 locked 30 pretty please?
(LA Noire does not count by the way)
GTA6 locked 30 pretty please?
(LA Noire does not count by the way)
alexandros
Banned
There is, I believe, a whole article accompanying the screenshots and discussing the issue, yes?
They can't even get a remaster to a locked 30, so the answer is most likely never.
RDR felt like it was at about 20fps in Armadillo.
The event is random, hence the reason the PS4 version might have a lower FPS is because it's doing something the XBO version isn't.
Now watch the whole video, in particular 1:10 to 1:30 (IIRC) there's 2 high speed car moments which are 'scripted' (same cars) and those both perform better on PS4.
This proves little other than 'in some random scenarios XBO has a better frame-rate' and that is likely not to be due to any CPU bottlenecks but more likely because it's doing more work than the XBO version.
nowheretorun
Member
It is more that while the 5 FPS difference is show the fact that you cannot recreate the scene in an exact manor on both consoles means it is not really informative. To be informative he would need to do the run say 5 times on both consoles, drop any outliers and average the remaining results to give you a true FPS difference in that scene. By doing the 5 runs and dropping outliers you get a more accurate feel for the true difference as while the scene will not always be the same over time it will converge on the true framerate difference.
Deaf Spacker
Memb
Yes but some people are freaking out saying that it's not locked to 30 FPS, it is impossible to lock a game like GTA even if you have an absolute beast as there's always something that can bring it down, especially if you go nuts.
This doesn't mean that current gen consoles are underpowered for the price you pay.
alexandros
Banned
I think Digital Foundry know how to benchmark a console game properly, I have no doubt they were scientific about it.
but even then we don't know if the XBO has less traffic (cut-back like the foliage to make it run better)
again, why would the other examples all have PS4>XBO
Did you have read the post above? We are discussing from days how this 'analysis' keep its conclusion from the...nothing. Because there isn't any kinda of evidence of those mystical advantages with the misery 9% of upgrade in the cpu speed, especially looking in the pc counterpart. It's just a stupid speculation started with Unity without any concrete prove.
alexandros
Banned
But Ubisoft developers themselves said their game was CPU-bound.
In the case of Mr Leadbetter the text of the article vs the video leads me to believe otherwise. Now he did say there were multiple runs and if that is the case a nice chart to show the average of all of those runs through that section would be nice.
alexandros
Banned
Did AMD have an SoC with something better than Jaguar cores back then? I can't remember.
they also said due to the NPCs and then later said that wasn't the case (or words to the effect of)
What the... who said otherwise?
Exept a gpu and a cpu does different things, so you can't really compare 1 gpu flop to 1 cpu flop. If you could why have a cpu at all? They would just drop the cpu altogether, put a titan and a lot of memory in there and have a super powerful console.
alexandros
Banned
Still, it is a recorded quote from a third party and as such it is worthy of discussion.
Again, this is all perhaps right but not necessarily. We just don't know, no matter how hard we try to find reasoings for whatever opinion we have.
We don't even know how the scheduler in these new consoles work. We know that only 6 of 8 cores are used but are they locked to certain cores? Does this change? This would have immidiately implications of how cache gets invalidated lines, for example. And this is only one uncertainty we have. This is why I am saying we need profiling data - without it, _nothing_ can be said for sure.
DeFiBkIlLeR
Banned
Yes, in terms of CPU performance (not GPU)...but they were desktop parts and they were outside the heat and power constraints required for the consoles.
It's funny to see you haven't read a single word in this thread because I want to see just a single post who tries to presume the shit you claimed here.
How do you explain AC:Unity then? I don't think all devs making AAA games will develop new engines
krypt0nian
Banned
Leadbetter.
alexandros
Banned
I don't get it. Is that supposed to mean something? I'm missing some context here.
DeFiBkIlLeR
Banned
Just another of the many DF haters on NeoGaf.