Digital Foundry Article Technical Discussion Archive [2015]

[Globalisateur]

Confirms the issues I have in the city...again my question, can the streaming problems in the city be improved with a SSD?

Yes, this post shows that PS4 has trouble to load textures and stuff in the big city compared to the XB1 version which could be related to the higher number of >33ms frame-time values than the XB1 game (~3 times more).

But those saying, notably in the Gaf thread, that the XB1 version runs much better than the PS4 version is total bullshit and hearsay directly taken from the DF debatable conclusion.

During the important and very frequent combat scenes, even taking into account the odd camera speed difference of the second combat scene, both versions objectively perform very similarly and in some cases the PS4 game performs even slightly better, still running at 44% higher resolution...

And finally it was the same in the big cities of GTA5 where the XB1 had the only advantage in junctions (where the streamed loadings are usually done). But in the end, the framerate has being totally patched on PS4 in those areas. But the others scenes where the PS4 game has always had the advantage (GPU taxing scenes) have never being patched on the XB1 game...

 

[iroboto]

Yes, this post shows that PS4 has trouble to load textures and stuff in the big city compared to the XB1 version which could be related to the higher number of >33ms frame-time values than the XB1 game (~3 times more).

But those saying, notably in the Gaf thread, that the XB1 version runs much better than the PS4 version is total bullshit and hearsay directly taken from the DF debatable conclusion.

During the important and very frequent combat scenes, even taking into account the odd camera speed difference of the second combat scene, both versions objectively perform very similarly and in some cases the PS4 game performs even slightly better, still running at 44% higher resolution...

And finally it was the same in the big cities of GTA5 where the XB1 had the only advantage in junctions (where the streamed loadings are usually done). But in the end, the framerate has being totally patched on PS4 in those areas. But the others scenes where the PS4 game has always had the advantage (GPU taxing scenes) have never being patched on the XB1 game...

Though, if I am not mistaken from skimming all the DF articles, PS4 has in many cases for multiplatform games has had more challenges with Pop-in/ stream in performance vs XBO and PC.
Streaming performance likely has heavy influence from the memory architecture.
Again, to not turn this into another AF missing on PS4 debate, the game and the settings are not necessarily reflective of the performance of the hardware. In the end it's quite impressive that they got the game to perform so close together, that is quite a feat.
GAF is a fun place to read, I am highly entertained by how the community reacts to each other. But in the end they are a different type of forum from here, some of the rules that they apply to posting wouldn't apply here. That said, they are certainly effective at creating a burning platform.
We should try to keep it technical here. No article is going to be able to represent the whole game. Unless you are interested in frame capturing the whole game from beginning to end on Xbox and PS4 I would not lose sleep over it.

 

[function]

So the blindingly obvious happened.

With the 30 fps cap in place the Xbox One version of Witcher 3 doesn't drop to PS4 levels. It remains higher at stress points, sometimes significantly higher. It's not close.

I love it when the entirely predictable happens.

 

[goonergaz]

Though, if I am not mistaken from skimming all the DF articles, PS4 has in many cases for multiplatform games has had more challenges with Pop-in/ stream in performance vs XBO and PC.
Streaming performance likely has heavy influence from the memory architecture.
Again, to not turn this into another AF missing on PS4 debate, the game and the settings are not necessarily reflective of the performance of the hardware. In the end it's quite impressive that they got the game to perform so close together, that is quite a feat.
GAF is a fun place to read, I am highly entertained by how the community reacts to each other. But in the end they are a different type of forum from here, some of the rules that they apply to posting wouldn't apply here. That said, they are certainly effective at creating a burning platform.
We should try to keep it technical here. No article is going to be able to represent the whole game. Unless you are interested in frame capturing the whole game from beginning to end on Xbox and PS4 I would not lose sleep over it.

I would have thought HDD could affect that depending on the game.

So the blindingly obvious happened.

With the 30 fps cap in place the Xbox One version of Witcher 3 doesn't drop to PS4 levels. It remains higher at stress points, sometimes significantly higher. It's not close.

I love it when the entirely predictable happens.

Did you not read what Globalisateur said - he's suggested that the playing field was not level and there were examples of PS4 performing better anyway - more a draw (in gameplay).

 

[function]

Let's balance out the playing field by discounting the parts where the Xbox one is showing a clear lead. Cutscene? Doesn't count. On a horse? Doesn't count. Moving the camera? Doesn't count.

Same old same old. Apply the template to AssCreed, CoD, whatever.

My main point though was that the idea that a 30 fps cap for Xbox one would see performance drop significantly in none capped section, and that the PS4 was actually performing better, was always unlikely.

Now you will notice that those goal posts have evaporated and new ones have appeared.

The real reason that the playing field isn't level is because the PS4 is rendering at a 44% higher resolution. Because 1080p.

 

[iroboto]

I would have thought HDD could affect that depending on the game.

Eh. 5400 rpm drive in both. I'm not seeing a reason for there to be such a difference between the two platforms if the HDD is to blame.

 

[psorcerer]

The real reason that the playing field isn't level

No, the real reason is the game performance. There is nothing there, graphics-wise, that justifies these high frame times, on any platform.

 

[Cyan]

Yes, and frame times will vary when frame rate is unclocked (just like they will on console). But that's not what was said above. As a reminder, Cyan said the following:

"When I play PC games the fps chart moves everywhere, up, down, down, up, up, down, in front of my eyes. If we were talking about a range of 3 fps or so, that makes no difference, but sometimes you get 90 fps and then after a moment 25 fps in some situations."

To which you responded:

"That's the nature of DX11 (or 10, or 9, things started going downhill approx. from the time that pixel shaders were introduced, for obvious reasons),"

Even if you weren't, Cyan was certainly referring to the average framerate and not individual frame times. I won't deny that some (very badly optimised) games will display strange, seemingly unjustified average frame rate fluctuations (CoD:Ghosts springs to mind as the obvious example). But this is absolutely the exception and not the rule as implied above, and can also apply to console games occasionally as well.

I've seen your video and I don't see major framerate drops, that's true. Even so, you locked the framerate, and most people can't do that in their GPU control panel. I know I can't for instance, but I can enable Vsync, at least.

Good examples of the always changing behaviour of the framerate can be seen in the most recent Digital Foundry video:

AC Unity is one of the worst offenders, the framerate can be at 100+ fps and then run at around 60 fps, and The Witcher 3 goes from 90+ fps in a scene to 60 fps under a split second afterwards.

CoD: AW is running at 210fps in one scene and suddenly decreases to 130 fps, and I mean it. Or from 68 fps on the R9 290X to 150 fps...

http://www.eurogamer.net/articles/d...e-witcher-3-patch-103-fix-console-performance

Digital Foundry and patch 1.03 on Xbox One and PS4 framerate is a bit more stable on Xbox One the biggest difference is visible in cutscene.

I hope they will continue to improve performance on both sides.

For a game like this, I expect great improvements over time. IF you take into account the incredible complexity of this game, I am impressed how there aren't many bugs -although they exist.

 

[Cyan]

On a different note, given the repercussion of Digital Foundry, I think it would be a good idea if they analysed non AAA titles all the time, and promoted some smaller budget titles that deserve all the praise, technically wise, or even if they were not that good technically, to help make those games known.

I mentioned that in this Beach Buggy Racing thread, which is not only a very fun game, but it's technically impeccable, and imo it deserves some attention.

https://forum.beyond3d.com/threads/...fps-original-fun-skids-like-mario-kart.56905/

Either that or an Eurogamer article roundup of small/medium budget titles could get the job done.

You wouldn't have to stop delivering the coverage of AAA titles, but will be relieved seeing that those articles don't take the same amount of flak, it would enrich Digital Foundry and it wouldn't suppose a heavy effort to take them done. People would start to like DF more, too, because most of those titles are usually not meant to be scrutinised like AAAs. Just my 2 cents...

 

[chris1515]

http://m.neogaf.com/showthread.php?t=1054862

They are working on a new patch with performance improvement for all platform and they speak about framerate of PS4 version.

For all platform: Improved camera smoothness

 

[Globalisateur]

...

The real reason that the playing field isn't level is because the PS4 is rendering at a 44% higher resolution. Because 1080p.

You clearly don't want to understand here. It's is a draw particularly when the resolution matters, here during combats in the countryside. The XB1 is only performing better here in the 2 instances when resolution doesn't matter:

- In the cutscenes, the PS4 engine is capped at 20fps each times the engine drops too much (we don't know what too much means here unfortunately), so the drops (even when very short at ~25fps) mean absolutely nothing as we don't know if it's enforced by the engine to reach 20fps or because of a CPU / GPU bottleneck.
- In the very big city, there is a constant loading of new assets (as proven in the constant late, compared to XB1, loading of textures and characters seen in my previous link), so the drops are not again directly related to resolution whatsoever but more of an API (and hardware?) situation causing a streamed loading bottleneck seen in many others games, particularly on PS4.

What this video proves is that, when we are sure only the resolution matters, it's is technically a draw between both platforms. If you want you can say overall that the XB1 performs better in this video, yes, but you can't say it's because of the 44% increase of the resolution, based of this video, quite the contrary in fact.

 

[Arwin]

Yes, the cutscenes are 1080p on XboxOne too, right? The PS4 uses some kind of 20fps vsync (i.e. go too much under 30, then lock at 20), that I expect they will drop in the next update. We'll see.

EDIT: I should say 20fps capped framerate, should know better by now

 

[goonergaz]

Eh. 5400 rpm drive in both. I'm not seeing a reason for there to be such a difference between the two platforms if the HDD is to blame.

Sorry, I was under the impression the XBO had a hybrid HDD and the PS4 a slow one...my bad.

 

[Shifty Geezer]

You clearly don't want to understand here.

There are two comparisons being made. One is specifically the framerate. The other is the game overall. For those like Cyan, XB1 'wins' because framerate. For others, PS4 wins because resolution. For the rest of us, XB1 has a framerate advantage and PS4 renders at a higher resolution and DF has done its job and we can move on to look at another game, maybe discussing the choices of devs as to framerate vs res (oh noes, not again! ).

 

[goonergaz]

Did I misread Globalisateurs comments? I thought (other than cutscenes) there was a case for pretty much minimal differences (if any) in framerate? Naturally if the camera is being used it would have been nice to closer match for a fair comparison?

 

[Globalisateur]

Yes, the cutscenes are 1080p on XboxOne too, right? The PS4 uses some kind of 20fps vsync (i.e. go too much under 30, then lock at 20), that I expect they will drop in the next update. We'll see.

Yes and unfortunately we won't be able to fairly compare very GPU taxing scenes like the swamps because like the cutscenes, there is an annoying 20fps cap even during gameplay on the PS4 version, when the XB1 game hovers at ~25fps, the PS4 game is automatically capped at 20fps...

 

[Arwin]

No, the real reason is the game performance. There is nothing there, graphics-wise, that justifies these high frame times, on any platform.

What do you mean exactly? That the graphics are badly programmed, badly optimised, or that they should have prioritised keeping frame budgets more? And what do you base that on?

 

[Shifty Geezer]

He means if the devs had ignored the CPUs and written the entire games on the GPUs, it'd be running 60 fps.

 

[iroboto]

Sorry, I was under the impression the XBO had a hybrid HDD and the PS4 a slow one...my bad.

Lol I wish the XBO had a hybrid drive. Oh one can dream. The current load times are atrocious on XBO.

Upgrading to a 7200 rpm external drive has quite a bit of improvement. But I'm too lazy lol

 

[iroboto]

Yes and unfortunately we won't be able to fairly compare very GPU taxing scenes like the swamps because like the cutscenes, there is an annoying 20fps cap even during gameplay on the PS4 version, when the XB1 game hovers at ~25fps, the PS4 game is automatically capped at 20fps...

I'm reading that PS4 is double buffered? Which is why it is dropping to 20 FPS lock. Not a clue what that means.