Cyborg
Member
XB1 = grass theft auto
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
Digital Foundry: GTA V PS4 and Xbox One compared in new frame-rate stress test.
- Thread starter Hanmik
- Start date
XB1 = grass theft auto
Fixed it for ya.
But the ISA today is only an additional layer of abstraction for an underlying risc implementation. Also although the ISA on x64 has grown over decades only a subset is used so writing programs or parts of it in assembler isn't that more complicated than other ISAs.
holygeesus
Banned
Have any X1 version owners, done any comparisons between running the game on the internal and an external drive? I'm wondering if it helps with asset streaming or other aspects of the game, especially those of you who use 7,200rpm external drives.
Cheers.
Cheers.
It's the smoothest running GTA ever released on a console.
Wouldn't achieve that much.
Would should achieve more is huma and the APU strengths being used fully which currently isn't the case as the cpu and gpu are still being treated as separate entities like on current PCs.
*Edit* Damn phone, accidental double post.
interfreak
Member
This actually sums it up almost perfectly - hopefully the PS4 'traffic stutter' can be patched.
Played my copy on PS4 today. Dear god that frame rate just dives as soon as you drive quickly in the city, it's awful - very quickly turns into a constant loss of frames staggering all over the place. How they thought this was acceptable in a previous gen port is beyond me.
It does seem to be to do with the speed assets are loaded, because if you jam the brakes on while in one of these areas, regardless of the amount of traffic around you panning the camera is 30fps.
It does seem to be to do with the speed assets are loaded, because if you jam the brakes on while in one of these areas, regardless of the amount of traffic around you panning the camera is 30fps.
interfreak
Member
Yeah I've learnt to drive a bit slower around the city, had really helped to reduce the frame rates issues.
I think it may be the smoothest GTA we'll ever get.
I have a feeling Rockstar's next GTA may be a little TOO ambitious and GTA's performance will return to how it usually does.
alexandros
Banned
I wonder why Rockstar didn't just lower the level of detail in busy environments.
serversurfer
Member
Well, yeah. Actually, I dunno about the Wii U, because I know nothing about its architecture, but I assume it has some capacity for GPGPU.
Now, as I was saying before, the lack of hUMA on most PCs would tend to limit its usefulness, because of all of the data transfer back and forth between the two processors. For example, you might have a function that takes 5 ms on the CPU, but only 1 ms on the GPU. That seems like a huge win, but if it takes you 2 ms to send the data over, and another 2 ms to get your reply, you haven't really gained anything. Basically, any gains you find need to outweigh the roundtrip on the data itself. hUMA avoids this by sharing data directly between both processors. Nothing needs to be passed around, so you get the full benefit of any gains you find.
Regarding XBone, it's obviously GPGPU-capable, but I don't know if it's actually a hUMA architecture or not. I've seen plenty of people claim that it is, but I've seen some technical info that seemed to indicate it wasn't. So, maybe, maybe not. Regardless, PS4 has additional hardware to further facilitate GPGPU which the XBone lacks, so not all techniques will translate.
So yeah, all of the current major players can make use of GPGPU, but some are better at it than others. Supporting PC would definitely hold things back, or require fairly different engines, at least. Supporting XBone will also cause some porting issues, but perhaps not as bad as porting to PC would, if it is actually hUMA.
Well, it's obviously different, or it would be the same.
That said, I don't think it's terribly difficult to implement, though like with anything, I'm sure some find it easier than others. It's not Dark Magic that only three people in the world actually understand, if that's what you're getting at. Yes, it's really possible to use this stuff, if you're so inclined.Actually, Sony's tools make this process comparatively easy. First, they already have a mature scheduler they adapted from the SPURS system they designed for the Cell's SPUs, since they operate in much the same fashion as GPGPU work. Even better, their compiler will give you two versions of any given function you write; one to run on the CPU, and one to run on the GPU. So you simply load up a dummy data set, race them, and implement the winner.
alexandros
Banned
Thanks for the explanation!
Panajev2001a
GAF's Pleasant Genius
If compilers were perfect, LHS and slower micro-coded instructions would have not been a problem on the previous generation consoles either and incidentally IA-64 would have ruled the world
.I think x86 was a smart choice for a lot of reasons, but I think we are focused on how we can make awesome CPU's despite the ISA they use, when your manufacturing process and CPU architecture chops are super state of the art, with ISA does not matter. Also, as I said, when you go massively multi-core all that CISC-y front-end's cost does count, a lot. X86 shines in big fat cores where the front-end is a minuscule part of the overall chip and Intel can turn the rest of the micro-architecture to 11
.What we are robbing ourselves of is the ability to conceive something beyond that and how that thing could be if it had the same technology and R&D pushing it. Even at Intel ll the other projects to replace it got eventually sidelined. I am still one of the guys who says Intel's own EPIC/IA-64 (Merced/Itanium) did not get a fair chance by being saddled with a much older manufacturing process compared to x86 CPU's. Also, their biggest chance to knock it out of the park was ruined by internal political ploys (Elbrus 2K managed to kill it from the inside in a way
) after Intel spent a great deal of time and money to get the CPU design team that worked on Alpha's EV7 and EV8, perhaps two of the best CPU's ever done (well the EV8 never came out unfortunately ...). That team was working on a next-generation IA-64 core, a much leaner single-banger speed demon multi-core chip (single-banger = each core would process one VLIW bundle at a time and not two)... Work was killed off, tons of people were rumored to have left as a consequence to that.Anyways, this is getting off topic, so I will stop here
.Strange, I'm not seeing this. In fact I'm enjoying the races because of how much more responsive this version feels compared to previous.
Windom Earle
Member
Yeah, it's bad. Especially when coming back into the city from smooth areas. This was the one thing I wanted fixed and they fucked it up. Waiting for patch and the definitive edition (PC).
Yeah, It gets even more noticable when you return from the city after driving around further up north. Raced through the city with Franklin on a fast bike, and it was chuggin most of the time.
Mr.Fletcher
Member
Feels great to me, I have no complaints. There's the odd drop here and there but nothing severe enough to impact enjoyment in my opinion.
Lol, I do the same. Feels strangely ..normal. #speedlimit
Stare-Bear
Banned
Quite disappointed actually with the PS4 version actually. A lot more stuttering while driving around in Los Santos then I expected... Also happens when you drive under a overpass. Hopefully they can patch it...
Same here, I'm not seeing the driving frame drops.
I guess ignorance is bliss, and it looks and plays great on the PS4.
AMD: PlayStation 4 supports hUMA, Xbox One does not
http://www.neogaf.com/forum/showthread.php?t=657221
You saw plenty of people who said that Xbone has hUMA architecture. Well, IIRC, it was just response from Xbox One dev because PS4 really has hUMA design.
This is his hilarious response :
A certified Xbox One developer took to Reddit to clear up the situation.
"Somebody had PM'ed me earlier linking to the Ars article on how the PS4 has a big advantage over the Xbox one due to hUMA. This was apparently said by some AMD marketing manager.
"I haven't heard of hUMA until today, so I went to look it up. The way I understand it is that in addition to having unified memory access (shared memory between CPU and GPU), which allows the GPU to read CPU memory, it is also a coherent cache system."
The dev added: "I remember reading something on this when I got my first alpha kit. I pulled up a couple of our internal white papers and it's pretty clear that this was the exact implementation in the Xbox One's memory system.
http://www.videogamer.com/news/xbox...huma_memory_system_just_like_ps4_say_dev.html
Well, before AMD's statement about hUMA in PS4, he knew shit about hUMA in Xbone.
Be glad that you don't see it for once you do see it, it'll stick out like a sore thumb and will be hard to go back.
I can see what DF were commenting on but it only manifests occasionally and is not a big deal imo.
DF word things to make them sound subtly worse than what they are for whatever reason that I don't want to speculate on. They made it sound like COD AW was bad on the PS4 because they didn't fully explain that the fps drops only happened when a grenade went off that filled the screen with a flickering digital red flash that meant you couldn't tell anyway.
I wonder if what we're seeing is people who only drive around town not having such bad issues because most assets are already loaded, however, when folk drive off to other areas the assets need to stream back in causing the noticeable drops.
As others have stated I've only seen minor non-game breaking frame drops...however I yet to venture out of the city, and during night, less traffic/junctions the PS4 version is butter smooth.
Again tho, all these 'XBO Driving vs PS4 shooting' should really watch the video,,,there's several driving scenes where PS4 outperforms XBO so it's not as simple as that...having said that their are no examples (from what I can see) where XBO outperforms PS4 during shooting (and the gaps are bigger and more consistent) so it's clear the PS4 performs better. There's even one junction where the XBO dips below PS4 to 25fps which again casts doubt on any true advantage over coincidence.
As others have stated I've only seen minor non-game breaking frame drops...however I yet to venture out of the city, and during night, less traffic/junctions the PS4 version is butter smooth.
Again tho, all these 'XBO Driving vs PS4 shooting' should really watch the video,,,there's several driving scenes where PS4 outperforms XBO so it's not as simple as that...having said that their are no examples (from what I can see) where XBO outperforms PS4 during shooting (and the gaps are bigger and more consistent) so it's clear the PS4 performs better. There's even one junction where the XBO dips below PS4 to 25fps which again casts doubt on any true advantage over coincidence.
For the same reason they didn't on ps360.
Pretty sells better than stable.
UnrealEck
Member
Yes, they could do with lowering some of the graphics a touch.
alexandros
Banned
IAmRandom31
Banned
My copy came in last night and I was quite shocked at the IQ.
Expected so much better. The shimmering is still god awful, just like it was last gen. It is cleaner yeah, but man .....
Expected so much better. The shimmering is still god awful, just like it was last gen. It is cleaner yeah, but man .....
I really dont get why we cant have a few minor options, even if it was really basic 'performance vs pretty'
I honestly don't think it's the detail, it seems to be when it's having to quickly load in more building models. Park up directly in the middle of a place which was chugging and panning the camera shows nearly zero drops, while cars constantly appear and drive past you in lesser or greater numbers.
JohnnyFootball
GerAlt-Right. Ciriously.
I believe Sony said it's possible. But it has to be taken into consideration what can be gained.
At most you would likely get a 20% in clock speed 10% being more reasonable.
But how much in terms of performance is really gained and is the additional heat being produced going to be properly accounted for?
Because the history of GTA games really shows, that Rockstar puts emphasis on playability and a stable framerate.
The history of GTA sales data also shows that Rockstars customers put an emphasis on a stable framerate.
To be clear: No one gives a fuck, not the devs, not the customers. Why change?
EDIT:
Oh, not to forget: The media doesn't give a fuck too. How many reviews mentioned these shortcomings? Right. Most of these journalists don't even know what framerate is.
GribbleGrunger
Dreams in Digital
I don't know if this has been posted but it's interesting that his conclusion is the complete opposite when it comes to framerates:
https://www.youtube.com/watch?v=O9seTLRgMoI
According to this, the PS4 has:
Slightly better foliage (I think he's being kind)
Slightly better lighting
Slightly better performance
https://www.youtube.com/watch?v=O9seTLRgMoI
According to this, the PS4 has:
Slightly better foliage (I think he's being kind)
Slightly better lighting
Slightly better performance
GribbleGrunger
Dreams in Digital
So basically what DF have done is take the odd occasion when the XB1 performs slightly better and made that the centre of their argument that the XB1 performs better 'overall'. Interesting ...
This is obviously why the DF video doesn't seem to mirror what the text is saying.
These guys need to set up a "sub site" which gives us face offs ala DF.
More indepth and clearly unbias.......for now.
GribbleGrunger
Dreams in Digital
He's worth following on Youtube. He's a tad pompous at times but he does great breakdowns of tech too.
GribbleGrunger
Dreams in Digital
No. It appears that the majority of the time the PS4 version runs slightly better but there are occasions when the XB1 version runs slightly better. So he's taken the odd occasion when the XB1 version runs slightly better (by the odd frame or two) and suggested that it trumps the PS4 version running better overall. I disagree with that completely. To explain better: The PS4 version may run 1 - 2 frames better 70% of the time but the XB1 version runs 3 - 4 frames better 10% of the time. Conclusion: The XB1 version performs better overall. That make absolutely no sense at all. The PS4 version runs better overall but the XB1 version runs better occasionally.
It sounds like a very biased approach to me. This is why the videos don't match the text. If they continue with this in the future I'm sure they'll be even more selective in the sections they show, which will make the whole thing pointless.
I see.
I've heard about this Leadbetter guy. I read the writing and some of it appears to me as subtlety biased. Honestly, affects credibility of DF as a whole.
robertsan21
Member
so its time for a new unbiased website to do these analysis for us. what website would that be?