• 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.

Hanmik

Member
They have updated their test with this segment..

UPDATE 20/11/14 4:03pm: The final analysis is in. We've put together a wealth of GTA 5 clips culled from capture sessions spanning over 20 different missions, concentrating on areas where the title's 30fps target frame-rate faces its toughest challenges. In comparing PS4 and Xbox One, the end result is fascinating; it shows that while there are some scenarios that prove equally challenging for both consoles, specific areas can see one platform out-performing the other. On a native 1080p title like this, we might reasonably expect PlayStation 4's more powerful GPU to demonstrate an unassailable advantage across the game, but the reality is a little more complicated.

Travelling at speed through city junctions - such as Strawberry Avenue - turns out to be a particular Achilles Heel on PS4, with sustained drops to around the 24-26fps range at their worst. It's a situation that also applies to Xbox One, but to a consistently lesser degree when testing under the precise same conditions (and borne out in repeated tests). Put simply, in areas where junctions are stacked up one after the other, the Sony platform suffers more prolonged frame-rate drops when burning through traffic with your foot jammed to the floor. It's an interesting, recurring scenario that points to a CPU bottleneck, where Xbox One's increased clock-speed has an advantage when racing around these busy sections.

Another potential cause could be hard drive access, of course. GTA 5 is very intensive in terms of background streaming, a state of affairs that made the digital download versions of the last-gen version less desirable than the physical releases (where streaming bandwidth was shared between disc and drive). To lessen the impact of any streaming bottleneck, we installed the PS4 version onto an SSD for a re-test - the results showed no improvement.

However, while Xbox One enjoys an advantage during high-speed races in packed junction areas, it's clear the platform faces its own particular challenges. It drops a few frames during downtown driving outside of the problem junction areas, whereas PS4 remains solid. On top of that, we've already seen that Rockstar has made some visual cutbacks to its complex outdoor rendering, presumably to accommodate Xbox One's less capable graphics hardware, but in certain areas where cuts haven't been made, performance can dip.

Complex effects work, usually involving transparency effects (explosions etc) can also cause visible, sustained hitches in performance. Explosive missions with Trevor in Los Santos' outskirts show the biggest pitfall on Xbox One - with 24fps held at length during one shoot-out. Meanwhile, the PS4 has no issue with these segments, sparing one or two dropped frames just as alpha effects appear on-screen.

It's an interesting result overall, leaving us with different performance profiles on each platform. It's an unmistakable advantage for Sony's platform when it comes to shoot-outs, but on the other hand, the Xbox One typically avoids the heavy mid-20s frame-rates that can kick in during high-speed chases.

more at the link

http://www.eurogamer.net/articles/d...er&utm_medium=social&utm_campaign=socialoomph
 

Frumix

Suffering From Success
Grass has never been this cinematic.

No, seriously am I missing something here? Why is a last-gen up-port enhanced it may be chugging sub-30? Devs stop feeding this slop to us and demanding 60 dollars for it.
 

LTWood12

Member
Weird, and honestly disappointing to see a game from last gen (albeit maybe the biggest game from last gen) still not hit a rock solid 30 fps. Really wish they would cut back on the effects to keep the frame-rate stable. At least give us the option to do so.
 

Marlenus

Member
Is it possible for Sony to increase the CPU clock speed via a firmware update ?

Unlikely, they will have been binned and tested based on the current clock speed, at the current voltage settings with the current cooler. While 90% of consoles may be able to support the increased clock speed without upgrading the cooling system or increasing the voltage there will be a few that will not because the particular chip only just passed the tests at the desired speed bin.

I still find it funny that they are attributing a > 9.4% FPS advantage to the CPU alone. It is likely part of the reason but when you are comparing 24 FPS to 28 FPS you are talking about a 20% gap so there must be something else going on too.
 
Weird, and honestly disappointing to see a game from last gen (albeit maybe the biggest game from last gen) still not hit a rock solid 30 fps. Really wish they would cut back on the effects to keep the frame-rate stable.

Well it's probably a very CPU heavy game, so it's not that strange.
 
Weird, and honestly disappointing to see a game from last gen (albeit maybe the biggest game from last gen) still not hit a rock solid 30 fps. Really wish they would cut back on the effects to keep the frame-rate stable.
In PS4 case, it is a CPU bottleneck and in XBO case, it is a GPU bottleneck.

We never really got many solid 30 fps on the previous generation so don't think this will be the case in this generation.
 

StuBurns

Banned
Unlikely, they will have been binned and tested based on the current clock speed, at the current voltage settings with the current cooler. While 90% of consoles may be able to support the increased clock speed without upgrading the cooling system or increasing the voltage there will be a few that will not because the particular chip only just passed the tests at the desired speed bin.

I still find it funny that they are attributing a > 9.4% FPS advantage to the CPU alone. It is likely part of the reason but when you are comparing 24 FPS to 28 FPS you are talking about a 20% gap so there must be something else going on too.
They did it for the PSP.
 

ElTorro

I wanted to dominate the living room. Then I took an ESRAM in the knee.
They did it for the PSP.

Only because the PSP was shipped in an underclocked state. The original plan has always been to run it's CPU at the clock speed that it ended up with.
 

-griffy-

Banned
So they both perform solidly most of the time but have notable drops in certain, different circumstances.

In before people try and claim one version has a clear, undeniable advantage over the other in regards to framerate.
 

Marlenus

Member
They did it for the PSP.

The only way that was possible was if they binned the chip for the higher speed in the first place but only enabled the lower speed to start off with. In effect they would have underclocked the chip. There is no other way to guarantee that it would work at the higher speed grade for all the existing units.
 

Sorral

Member
I still find it funny that they are attributing a > 9.4% FPS advantage to the CPU alone. It is likely part of the reason but when you are comparing 24 FPS to 28 FPS you are talking about a 20% gap so there must be something else going on too.

They mentioned that parts of downtown is 30fps on PS4 while X1 drops frames at that same area.

The CPU clock reasoning isn't good because of that. Just screams that this could use some more optimizing.
 

DBT85

Member
They did it for the PSP.

PSP came out downclocked.

The better comparison is the Xbone which would have gone through exactly the same process as the PS4 chips. Only Sony know if they c an turn it up 10% like MS did.

I imagine their other option is to make compute as easy to implement as possible for devs.
 
This is gonna be a good thread.

Did I say good? I meant bad.

I'm honestly happy to see both systems performance near matching. Although hopefully devs start taking advantage or each systems strengths instead of meeting in the middle further on in the generation
 

Percy

Banned
They say that there have been visual cutbacks on the Xbox One version in the outdoor parts of the game and that the Xbox One frame dips are shorter during high speed movement in certain areas... and they don't think to address whether these two things could be related?
 

Marlenus

Member
So they both perform solidly most of the time but have notable drops in certain, different circumstances.

In before people try and claim one version has a clear, undeniable advantage over the other in regards to framerate.

I dunno. Dropping frames in one specific area of the city is one thing but dropping frames in all shoot outs is another.
 

KaiserBecks

Member
They didn't lock the frame rate to 30 on either versions...?

Sigh

I'd really love to know who started this myth. You can't "lock" a frame rate. You can cap it so it doesn't go over 30, but it can always be less than that. Especially in a game that isn't heavily scripted.
 
So PS4 if you want to spend your time in the countryside and XB1 for the cities.
Glad that current gen gives us decisions like these.
 

Marlenus

Member
PSP came out downclocked.

The better comparison is the Xbone which would have gone through exactly the same process as the PS4 chips. Only Sony know if they c an turn it up 10% like MS did.

I imagine their other option is to make compute as easy to implement as possible for devs.

MS would have had to retest all chips to make sure it could handle the increased clock speed. There is no other way to guarantee it will work.
 

Mononoke

Banned
Oh boy.. Can we all just get along? Both versions players will enjoy.

Sure but some people have the option to buy this for multiple platforms and want to get the most out of their money. They want to get the best possible version available to them of their favorite game.

I see nothing wrong with the tests nor do I think they are malicious in what they set out to do.
 

ElTorro

I wanted to dominate the living room. Then I took an ESRAM in the knee.
The better comparison is the Xbone which would have gone through exactly the same process as the PS4 chips. Only Sony know if they c an turn it up 10% like MS did.

I imagine their other option is to make compute as easy to implement as possible for devs.

The more likely variant for them would be to free an additional CPU core. Right now, only 6 cores are available to games. It's hard to see what the OS is doing with two cores, so maybe there is a reserve there.

I don't think that an upclock is possible. If it were, why wouldn't they just have set the clock speed accordingly right from the beginning?
 

Kilau

Member
Dat CPU secret sauce.

Yeah I could tell the PS4 was sluggish in busy areas. It doesn't seem to be very consistent, some areas where it's less busy it will lag and other areas it won't.
 

Trogdor1123

Gold Member
I was playing this for a few hours on my PS4 on Tuesday and I noticed frame drops right away. Not sure what was happening as it was a pretty simple area. Not sure the issue was though. Either way, it wasn't long or frequent.
 

The Llama

Member
They mentioned that parts of downtown is 30fps on PS4 while X1 drops frames at that same area.

The CPU clock reasoning isn't good because of that. Just screams that this could use some more optimizing.

Yeah, that's the impression I get after playing. Seems like there's a few things going on that just needed to be optimized a bit more. Maybe they'll fix it in a patch (doubtful, tbh).
 

Trogdor1123

Gold Member
What defines a CPU intensive game and a GPU intensive game? Is it just the way its coded or is there certain things that a person can look for? Serious question.
 

forms

Member
I am a PS4 owner, and one of those that just love roaming around in sp sandboxes. In a game where traversal by car play a pivotal part, however, getting drops to the 24 range is just not something I am interested in. If 30 is supposed to be good then I want 30.

I could accept it in a sandbox if you pile cars on top of each other and create some insane explosion or similar, but not while simply around.
 

Sorral

Member
I'd really love to know who started this myth. You can't "lock" a frame rate. You can cap it so it doesn't go over 30, but it can always be less than that. Especially in a game that isn't heavily scripted.

I meant stable enough to not have drops and stay at 30fps most of the time..

Why argue semantics/literal wording lol
 
XB1 CPU is not causing less frame drops. Come on.

I never imagined a tiny increase could be so massive.

Both versions drop frames below 30, and according to those videos, they do it quite often (esepcialyl the xb1 in that firefight). The difference is pretty minimal in the frames dropped when and how they occur.

The bigger story should be that this is dropping frames BELOW 30 at all.
 

hagege_s

Banned
The more likely variant for them would be to free an additional CPU core. Right now, only 6 cores are available to games. It's hard to see what the OS is doing with two cores, so maybe there is a reserve there.

I don't think that an upclock is possible. If it were, why wouldn't they just have set the clock speed accordingly right from the beginning?

Good question... You are probably right about the OS BTW, same thing with the RAM, with time and optimisation, hopefully it will take less and less "power".
 
Top Bottom