Performance Analysis: Just Cause 3
UPDATE: The Digital Foundry budget PC hands in a good experience at close to max settings.
UPDATE 2/12/15 9:00am: We're still working on our Face-Off coverage for Just Cause 3, but in the meantime we've updated the article with how the game runs on PC - specifically our budget computer, featuring a Core i3 4130, 8GB of 1600MHz DDR3 plus an Nvidia GTX 750 Ti.
Original story: Available to some ahead of time, initial reports on Just Cause 3 console performance haven't been particularly positive - with an Xbox One preliminary analysis causing much concern. Going into our testing, we couldn't help but wonder - is the console version of Just Cause 3 really in bad shape? After all, Avalanche's recent cross-platform work on Mad Max was exemplary. Additionally, with the pre-release focus on Xbox One, what of the PlayStation 4 version?
We managed to grab the game ahead of release from a local retailer, and our analysis is based on both releases patched up to version 1.01. In terms of core image quality, we can confirm that the PlayStation 4 version is indeed running at native 1080p, while Xbox One operates at 900p. Initial comparisons of art assets, effects and lighting suggests that both versions are operating with the same visual feature set. Avalanche clearly felt no need to adjust GPU-intensive elements such as shadow quality - in this respect, both look equivalent to the high setting on the PC game.
The end result is that both editions of Just Cause 3 look very similar, but PlayStation 4 has the edge visually with a crisper image, meaning that the real issue is indeed one of performance. Both versions target 30 frames per second and there is something akin to an adaptive v-sync solution in play on both versions - but like a lot of games utilising this technique recently, tearing is limited to the very top portion of the screen. In motion, Just Cause 3 looks and feels much like a v-sync title - but in restricting tearing to the top of the screen, Avalanche hopes to lower latency and reduce stutter with a slightly more elastic approach to the game's rendering budget.
What's clear is that neither version of Just Cause 3 is able to sustain a locked 30fps, but it is indeed the Xbox One release that offers a noticeably lower level of in-game fluidity. In our tests, we've compared both cut-scenes and gameplay, and our lowest recorded drop occurs on Xbox One, where a massive explosion sends us plummeting to 20fps, while PS4 plateaus to 24fps in entirely different circumstances - with CPU power the likely culprit for bottlenecked performance there.
In between shoot-outs, both editions mostly hold at 30fps, but once things kick off, it's clear that Microsoft's machine struggles to a more noticeable extent. It's a little disappointing that neither version of the game can stick to its target frame-rate - high tempo combat with lots of performance-sapping alpha transparencies cause issues on both machines - but it's clear that PlayStation 4 has the edge in sustaining target performance more consistently. What's particularly disappointing is that even in areas with little or nothing going on, Xbox One can still have issues - moving into a forest area and spinning around on the spot can cause the engine's performance to drop to the mid-20s, while PlayStation 4 retains the lion's share of its performance.
This is the kind of test we utilise in order to overcommit VRAM on PC graphics cards, but with the consoles employing unified memory, we were surprised to see this issue manifest in this way. Just Cause 3's open world is abundantly rich, and its distance detail levels are very close to the PC version running at its best, so it's clear that there's a lot of data to stream - we may well be looking at a memory bandwidth issue on Xbox One, or else it's a matter of storage.
Just Cause 3 is renowned for its explosive action and this is reliant on the fidelity of its physics model. Avalanche pushes the boat out there with the Havok middleware, but this does cause performance issues on both consoles. Blowing up a bridge with C4 explosives causes a mid-20s dip on both machines as the structure crumbles into its component parts. Bearing in mind that not a huge amount is happening from a visual perspective otherwise, it does suggest that the Jaguar CPU architecture of both consoles is being pushed to its limit.
And that's a bit of a shame really - the charm of this game is its all-guns blazing style of chaotic gameplay, and it's a little frustrating that these issues occur. And of course, what we have here for you today is an initial test - we're only scratching the surface of the game, and in an open world title like this, there's every chance that the engine may be stressed to an even greater degree later on.
So there's where we're at with Just Cause 3 right now. The plan is to investigate console performance in greater depth, and to stack up the visual feature-set on display here with the PC version. What we can say right now with a fair amount of confidence is that the GTX 950 paired with a Haswell Core i3 processor can easily run Just Cause 3 at 1080p maxed with SMAA anti-aliasing and 8x anisotropic filtering and stay well north of 30fps. Our 'go to' card for budget PC gaming - the GTX 750 Ti - does struggle though, dipping to a minimum 23fps even with shadow quality pared back to the high setting (though overclocking can help) [UPDATE 2/12/15 9:00am: This has increased to 26fps with the release of Nvidia's game-optimised driver, and half-rate adaptive v-sync does a good job of stabilising performance at console-level 30fps). These are value-orientated parts - a more powerful mid-range gaming PC should have few problems running this game well. We'll report back on that as soon as we can.
