One of the most noteworthy criticisms we've seen of L.A. Noire is that there are some serious frame-rate issues on the Xbox 360 version of the game, while the PlayStation 3 release is significantly smoother, in contrast to the previous Rockstar RAGE open-world titles.
It's fair to say there are performance issues in both releases, but there remains an undeniable perception that the PS3 game holds up better under stress. To test this out we carried out two different analyses: first, we compared the game running identical video, with an emphasis on combining the high polygon characters with resource-intensive exterior scenes. In the second test we compared gameplay from the same sections of the game.
There are several similarities in the way the game performs that seem to suggest that L.A. Noire uses an offshoot of the same RAGE engine that powered GTA IV and Red Dead Redemption, or at least that there was a great deal of technology shared with Team Bondi. The PlayStation 3 version of the game runs with v-sync, while the Xbox version has a small amount of tearing right at the top of the screen. The tearing is essentially unnoticeable and on many screens it would be tucked into the overscan area anyway. Regardless, this odd performance profile is exactly the same as GTA IV. However, Rockstar North's 360 code ran with an unlocked frame-rate while it is clear that L.A. Noire is capped at 30FPS on both platforms.
What we see is that engine performance is mostly like-for-like, but when performance drops, 360 does seem to come off worse. It may well be that the RAGE engine is triple-buffered on PS3 (effectively keeping a frame in reserve if engine performance dips), while the 360 version is double-buffered - displaying one image while drawing the next. If a frame takes too long to render, the engine effectively stalls until the next screen refresh.
The very first clip in the analysis seems to demonstrate quite clearly that in areas of sustained stress on the engine, the Xbox 360 version drops to 20FPS, while the PS3 code pumps out more frames, resulting in a smoother image. If this were to carry forward into gameplay, the result would much more noticeable as visual feedback to controller input would be significantly laggier on 360 compared to the PS3 game.
Once again, the first clip in the video throws up some curious results. We see that both versions suffer from some serious dips in performance, and it's interesting to note that physics-based actions (such as a breaking window caused by gunfire) seem to cause the 360 game some issues. On top of that, once again we see situations where frame-rate drops on PS3 are magnified on 360 - for example, in the on-foot chase clip just over a minute into the video.
We do see situations where the Xbox 360 version runs smoother than its PS3 counterpart, but this may well be down to the context of the gameplay at that moment (cars and pedestrians on-screen at that moment, for example). There's little in the way of evidence for any scenario which sees 360 consistently outperform PS3.
In terms of other elements of engine performance, it's fair to say that streaming of assets seems to be something of a mixed bag. We basically gave up comparing texture quality across the two versions because at any given point we can see higher resolutions on one platform compared to the other, or we see a mish-mash of different quality levels - higher resolution environmental textures on one version, but lower quality decals.
If we had to make a choice it seems that the PS3 version is perhaps more robust, but there's not really much in it. Installing L.A. Noire to the Xbox 360 HDD doesn't appear to make a whole lot of difference either, aside from the usual advantages in doing so: reducing noise and wear and tear on the optical drive.
The overall feeling you get from the game is that the assets are almost certainly like-for-like but it's difficult to shake the feeling that there are bugs in the engine that are stopping the game from achieving its full visual potential. At the time of writing a new patch has appeared that is said to address some bugs, but we have been hard-pressed to find any tangible difference in this regard as of yet.
In addition to the streaming performance, we occasionally see some differences in draw distance too. It's not exactly a regular occurrence, and in terms of regular gameplay you will struggle to notice the difference, but each time we have seen it, it appears to be the Xbox 360 game that is affected, either by missing objects in the far background or else with lower LOD models. However, once again it should be stressed that this only really comes to light in direct A to B comparisons - during the run of play it's not a massive issue and it's hard to imagine that it would affect your enjoyment of the game.