Call of Duty: World At War Performance Analysis

A couple of weeks back we took a good, long, hard look at Call of Duty: World at War on Xbox 360, PlayStation 3 and PC. Rather than repeat myself, you can check out my findings over at Eurogamer here, while the exhaustive range of 720p comparison shots can be viewed here.

Regular Digital Foundry readers will know that we researched the Call of Duty 4 engine beforehand. Our prediction at the time was that with the more intensive sequel, Xbox 360 would fall more often into the ‘perceptual’ 60FPS area (that is, dropping regularly from full-fat 60fps, but not really noticing the difference) whereas the PS3 version would cross that perceptual divide.

In the event we were probably half-right. The 360 game is running less smoothly than its predecessors, but both versions - by and large - still manage to look smoother (and most cases better) than their competition, all of which run, or aspire to run, at 30FPS. What is clear is that Treyarch has done a lot of tooling about to make the most of both platforms - better shadows on PS3, the odd tweaking of environmental objects on both concoles. But even after all this, it’s an undeniable that the PS3 version is running at a perceptably lower refresh rate than the 360 version for most of the time.

With the new Digital Foundry frame analyser, we can take a peek under the bonnets of both versions in a way hitherto unseen. The results are quite surprising - while frame rate averages back up Xbox 360’s palpable advantages, it’s also clear that there are some engine elements where the Sony console more than holds its own.

So let’s get on with the videos. It’s all pretty straightforward. A running frame rate (averaged over 30fps) runs in the top-left and top-right for 360 and PS3 respectively. The graphs show PS3 frame rate in blue, Xbox 360 in green. Due to our limited After Effects skillage, if both versions run at 60FPS, you’ll just see the one green line.

Nothing’s more important the first impression and with both Xbox 360 and PS3 versions of the game, we see a near constant 60FPS as the game’s first engine-powered cut-scene kicks in. It seems that particle effects give the PS3 version a touch of concern, but we’re looking at platform parity until the action moves outside and the explosive effects kick off. Xbox 360 handles this entire sequence at 60FPS with much aplomb, whereas the PS3 version starts to lag as soon as the first building is destroyed.

This clip is a good example of why an average frame rate is not the best way to ascertain engine performance. Xbox 360 averages 58.54FPS while PS3 comes in at a considerably less impressive 50.5FPS. However, as you can see from the way the clip unfolds, 360 struggles in a couple of moments whereas the PS3 stays pretty constant around the 50FPS area.

Here’s the clip that we used to first demonstrate the frame rate analyser back on the old Digital Foundry blog, only this time we see both Xbox 360 and PS3 in full effect. Whereas rendering the ocean appears to have a tangible impact on the PS3 code, the 360 remains solid at 60FPS until a combination of different effects kick in. The water spray and wobbling liquid-screen effect seem to have the most impact.

A fairly static scenario, but the load on the engine is still quite significant. Just one soldier opening fire adds enough load to the engine to cause a tangible (and now visible) dip in the frame rate on PS3, whereas the 360 code remains solid at full-tilt 60FPS throughout. There is not a single dropped frame here in the entire clip, versus 134 on the Sony console. On a side-by-side screen test, the difference is not really noticeable - there’s little in the way of movement that would really give away the disparity in refresh rate.

At first glance the PS3 version appears to be exhibiting lighting and shadows not seen on 360. In fact, this is a bug - the wall hasn’t been blown open yet, so there’s nothing to create this effect. The same shadow and lighting kicks in on 360 once Sergeant Reznov has made his triumphant appearance. It’s interesting to note that while the exploding wall and smoke effect hits refresh rate on both platforms, PS3 maintains the advantage overall in one of only two clips from the whole selection.

A section of the game that caused plenty of comment when the differences were first highlighted. While there are slight changes to the environment between the two versions throughout the game, this one is the most blatant, with the jungle vines pared back on PS3. There are other more minor changes, but the bottom line is that the reduction in detail puts PS3 back in the game against Xbox 360, already struggling itself under the heavy load. Once the attack begins, the video de-syncs, but we can see that once again the smokey explosive effect causes real issues for the Microsoft console - at its nadir it actually hits 26FPS - the lowest score from either version in any of these tests (though strictly speaking, not from like-for-like footage).

A swift look at the tank section of Call of Duty: World at War. Performance on 360 is marginally superior, but once again, the smokey alpha effect takes its toll. We also wonder whether it’s just coincidence that the PS3 version is pushing a lot more of it about and still maintaining much the same refresh rate…

In this fairly static screen, PS3 levels at a 50fps average pretty much across the entire clip. The Xbox 360 version on the other hand maintains its refresh rate with just six frames dropped thoughout the duration.

This final clip from the somewhat out-of-place aerial battling section shows the most dramatic lack of parity between the two versions in any of the clips analysed: the 360 game doesn’t drop a single frame throughout the whole video - PS3 on the other hand appears to be unoptimised on even the fairly simplistic sections where the player is running through the plane. The uneven frame rate at the climax of the clip is also intriguing.

In the final analysis, it seems clear that the divide in performance between PS3 and Xbox 360 on the CoD4 engine remains in the new game, but the increased load does take a toll on both versions. It’s also clear that there are elements that do impact 360 performance quite substantially and in the same situations, PS3 runs at an even, albeit generally slower, keel. By and large though, we’re limited to analysing scripted instances using the engine - the more random elements during gameplay make life-for-like comparison more difficult. In the majority of non-like-for-like frame rate tests, 360 maintained its lead - perceptually it is clearly the smoother game. But perception only counts for so much, and by choosing the clips we have, we’re seeing what the exact same stresses do to both consoles.

Update: Desirous of a full 720p streaming HD compilation of all the above clips? Look no further.