Performance analysis: Tomb Raider Definitive Edition

The reality of PS4 1080p60 vs. Xbox One 1080p30.

Two development studios handling separate next-gen ports offer two performance solutions. Digital Foundry reports.

The unscheduled, early release this weekend of Tomb Raider: Definitive Edition on Xbox One and PlayStation 4 allowed us to get something of a head start on our upcoming Face-Off coverage. Factoring in the controversy surrounding the frame-rate differential between the two versions, we prioritised performance analysis in order to get our impressions to you as quickly as possible - and the results are going to make for uncomfortable reading at Microsoft.

Reports suggesting that the PS4 version of the game operates at a slick 60fps while the Xbox One game languishes at the console-standard 30fps are verified - up to a point. Deeper analysis reveals that PS4 frame-rates can vary significantly depending on the effects work at play, resulting in variable controller latency and some on-screen judder. For its part, performance on Xbox One is palpably lower - massively so judged by the numbers alone, but the experience itself is more consistent overall.

First up, let's compare the two versions of the game running exact like-for-like footage via engine-driven cut-scenes. On Xbox One, we see a lock at 30fps, with occasional dropped frames shifting results lower. Activity is far more interesting on the PS4 side though, with frame-rate varying from anything between 32fps to 60fps across our sample, with elements such as TressFX, depth of field and transparent alpha effects hitting frame-rate in a cumulative manner. Note in particular how scenes that switch between the TressFX-enabled Lara and the rest of the cast can see sudden switches in performance.

We've added some additional metrics beneath the video, because as the game is apparently running unlocked, figures such as average frame-rate are actually meaningful - on the PS4 side at least. With 45fps reports filtering across the internet the suggestion is that the Xbox One version is unlocked too, though in our testing the game certainly acts as though it's capped at 30fps (a good test for this is looking directly at the sky with as little geometry on-screen as possible - here we still see a 30fps limit in effect) meaning that direct frame-rate comparisons from a pure 'benchmarking' perspective aren't really viable - especially when two entirely different developers worked on each version (PC veteran Nixxes for PS4, United Front Games for Xbox One).

For what it's worth though, comparing PS4 and Xbox One when the Microsoft platform dips under 30fps is perhaps the best indication we have of relative performance, and can see some significant differences - up to 20fps in some scenes. However, even with the 30fps cap in place (in theory limiting output), sometimes Xbox One moves closer to its sibling's performance level, when the PS4 engine is really under load.

"There's a clear PS4 performance lead, but the comparison is skewed somewhat by the fact that Xbox One frame-rates appear capped at 30fps, while the Sony platform runs unlocked."

Tomb Raider: Definitive Edition compared on PS4 and Xbox One across a number of the game's engine-driven cut-scenes. The idea here is that relative performance is measured by allowing the engine on each version to process the same content.

Cut-Scene Analysis Xbox One PlayStation 4
Lowest Frame-Rate 27fps 32fps
Highest Frame-Rate 30fps 60fps
Average Frame-Rate 29.98fps 53.36fps

Across the whole selection of clips, the effect of the Xbox One frame-rate cap is dramatic - with a 78 per cent frame throughput increase on the Sony hardware, where the engine is allowed to display a newly generated frame as soon as it is ready rather than waiting for the next 33.33ms refresh as is the case on Xbox One.

From synchronous rendering in engine-driven cut-scenes, we move on to actual gameplay, where we capture action from the same areas of each version. Clearly the results here won't represent exact like-for-like testing, but they do give us a better indication of how each version actually plays. There's a 71 per cent frame throughput boost on PS4 in these tests that owes much to the unlock, but a remarkable variation of 33-60fps across the run of play. For its part, Xbox One is clearly more consistent, but a lowest reported frame-rate dip of 24fps isn't great. [Update 20:20 GMT: and as the more eagle-eyed have noticed, we also have a registered dip to 18fps, where the game seems to momentarily freeze for 12 frames at 4:12 in the video - though this isn't indicative of general performance. For those interested, this post explains the confusion - in effect, the frame-rate counter is updating every half second - as you can see, the graph itself is registering 24fps. Apologies for the confusion, We'll get this fixed up in time for the full Face-Off.]

The overall feeling we get from the game is that two different developers handed in two different performance levels, and decisions were made on how best to work with the results. With the PlayStation 4 averaging at 50fps and often hitting 60fps, frame-rate was left unlocked, producing the markedly higher results you see in the tables above.

For Xbox One, we can only speculate, but we suspect that a lower overall performance resulted in even more noticeable judder were the game to remain unlocked - which would look really unattractive compared to the capped 30fps frame-rate we see in the final game. What's curious from our perspective is that United Front Games on Xbox One would have benefited from a reasonably straightforward porting process from the original PC DirectX 11 code since both platforms use the same API, while Nixxes would have needed to translate the original PC version across to the PS4's LibGNM API - not exactly a walk in the park based on this presentation from Ubisoft Reflections, who are handling the PS4 version of The Crew, ported across from the PC DirectX 11 codebase. Another development source we reached out to suggests that the DX11 'driver' for the Xbox One still requires a lot of work.

"If you've got it, flaunt it, appears to be Nixxes' approach to frame-rate on the PlayStation 4 version of Tomb Raider Definitive Edition."

The choice between PS4 and Xbox One comes down to unleashing the raw power of the console vs. limiting performance in the interests of a more consistent experience.

Gameplay Analysis Xbox One PlayStation 4
Lowest Frame-Rate 24fps 33fps
Highest Frame-Rate 30fps 60fps
Average Frame-Rate 29.84fps 50.98fps

In terms of the Tomb Raider gameplay experience, for all the many advantages the unlocked PS4 frame-rate brings across a range of scenarios, there are areas where the variation in performance is somewhat off-putting. Based on our initial few hours with the game, we would have liked to see an optional frame-rate limiter built into the PS4's display settings (similar to the option found in the BioShock games). This would have given those seeking maximum frame-rate the chance to exercise the muscle of the console, while those who prefer a locked, consistent, judder-free experience could have chosen that instead. Everyone's a winner in that scenario. But clearly, when Tomb Raider PS4 is running at 60fps - or close to it - the experience really is quite wonderful.

We've only just started to look at the Definitive Edition. Performance aside, there are still many different aspects of the game we're looking to cover - specifically, just how much of an improvement the new release is over the last-gen console game and whether it's worth upgrading, along with how the new release stacks up against the PC version running at its ultra and ultimate settings.

In terms of performance, we hope to have our consistency graphs finally up and running: our attempt to visualise judder and variable controller response, an area in which our current frame-rate graphs fall short. Look for our full analysis on the Definitive Edition a little later in the week.

Comments (473)

Comments for this article are now closed, but please feel free to continue chatting on the forum!