Although I didn’t spend much time playing Batman: Arkham Origins, I remember the game rather well after testing it on no less than 30 graphics cards and 20 CPUs. Arkham Origins appeared to take full advantage of Unreal Engine 3, it ran smoothly on affordable GPUs, though it’s worth remembering that Origins was developed for last-gen consoles.
This week marked the arrival of Batman: Arkham Knight, the fourth entry in WB’s Batman: Arkham series and a direct sequel to
2013’s Arkham Origins 2011’s Arkham City.
Arkham Knight is also powered by Unreal Engine 3, but you can expect noticeably improved graphics, in part because the PlayStation 4 and Xbox One have replaced the PS3 and 360 as the lowest common denominator.
Leading up to launch, we had high expectations for the game’s graphics after its recommended system specs were shared. As an Nvidia sponsored title, the specs failed to mention any AMD GPUs, but we were surprised to see the GeForce GTX 980 recommended for maximum quality graphics with the GTX 760 suggested for standard play.
Oddly, 3GB of VRAM was also called for even though most GTX 760 cards only sport 2GB, while the minimum specs demanded a GTX 660 with 2GB of VRAM.
Favoring one camp of GPUs is hardly the least contentious way to launch a new game, however it seems this release would have rattled cages regardless. Much of the PC gaming community has been angry about Arkham Knight’s performance with reports of constant stuttering that ruins gameplay. The issue is mentioned by users of both Nvidia and AMD cards, though it seems AMD folks are having the most trouble.
AMD has responded quickly with a new beta Catalyst driver that resolves some of the issues. However for AMD the problem is that like other recent releases such as The Witcher 3 and Project CARS, Batman: Arkham Knight is laced with Nvidia’s performance crushing GameWorks features including PhysX clothing and destruction on both PC and consoles. There are a few features that can only be enabled on Nvidia GPUs, such as interactive paper debris and interactive smoke/fog.
Some visual features like enhanced rain and enhanced light shafts can be enabled on all hardware, but overall the graphics menu is void of options. For instance, anti-aliasing can be turned on or off, with no mention of what AA mode is even used.
Enhanced rain means raindrops bounce off Batman’s cape realistically as he jumps, flies, glides, grapples and dives through Gotham City. As helicopters fly overhead, volumetric lighting creates what developers call “god rays” that shine down from the search lights.
Between GameWorks, the limited video options, and general lack of optimization, Arkham Knight’s performance paints an unusual picture. And it won’t be a pretty one. At least we hope it doesn’t run as poorly as Wild Hunt did with GameWorks enabled…
Arkham Origins had a benchmark mode, but it was so crude we preferred to benchmark using Fraps. Arkham Knight’s built-in benchmark seemed to work well enough that we initially opted to use it, making it easier for you to compare results. As you’ll see however, we wound up including Fraps testing as well for a more complete look at the game’s performance on today’s hardware.
Benchmarking was conducted at 1080p, 1440p and 2160p (4K) resolutions using the highest quality settings available to us. Enhanced rain and enhanced light shafts were enabled, while we didn’t test with interactive paper debris and interactive smoke/fog as they cannot be used with AMD graphics cards.
The game is locked at 30fps unless you modify a configuration file, which is among the top complaints by those playing on PC, signaling the issues that were to come.
Test System Specs
- Intel Core i7-5960X (3.00GHz)
- x4 4GB Kingston Predator DDR4-2400 (CAS 12-13-13-24)
- Asrock X99 Extreme6 (Intel X99)
- Silverstone Strider Series (700w)
- Crucial MX200 1TB (SATA 6Gb/s)
- HIS Radeon R9 390X (8192MB
- HIS Radeon R9 390 (8192MB)
- HIS Radeon R9 380 (4096MB)
- Gigabyte Radeon R9 290X (2048MB)
- Gigabyte Radeon R9 290 (4096MB)
- Gigabyte Radeon R9 285 (2048MB)
- Gigabyte Radeon R9 280X (3072MB)
- HIS Radeon R9 270X (2048MB)
- HIS Radeon HD 7970 GHz (3072MB)
- HIS Radeon HD 7970 (3072MB)
- HIS Radeon HD 7950 (3072MB)
- HIS Radeon HD 7870 (2048MB)
- Gainward GeForce GTX 980 Ti (6144MB)
- Gigabyte GeForce GTX 980 (4096MB)
- Gigabyte GeForce GTX 970 (3584+512MB)
- Gigabyte GeForce GTX 960 (2048MB)
- Nvidia GeForce GTX Titan (6144MB)
- Gigabyte GeForce GTX 780 Ti (3072MB)
- Gigabyte GeForce GTX 780 (3072MB)
- Gigabyte GeForce GTX 770 (2048MB)
- Palit GeForce GTX 760 (2048MB)
- Gainward GeForce GTX 680 (2048MB)
- Gainward GeForce GTX 660 Ti (2048MB)
- Microsoft Windows 8.1 Pro 64-bit
- Nvidia GeForce 353.30 WHQL
- AMD Catalyst 15.6 Beta
There are many interesting results at 1080p that we need to discuss. Starting near the top of our graph we see the GTX 780 averages a solid 64fps with a minimum of just 49fps while the R9 290 performs exceptionally well, delivering an average of 70fps and a minimum of just 51fps, arguably being the best value high performance GPU for playing Arkham Knight at 1080p.
The new R9 390 was just a single frame faster than the R9 290 while the R9 390X was just 2fps faster than the R9 290X and both delivered the same minimum frame result. The minimum frame rate seems to be limited by the Core i7-5960X processor, as the GTX 980 Ti would also drop down to just 53fps despite an average of 89fps.
Moving further down the graph we find the GTX 960 delivering playable performance along with the HD 7970 GHz Edition. The old GTX 680 also performed quite well, despite dropping down to 36fps at times.
It was interesting to find that for whatever reason the R9 380 and R9 285 would tank down to just 20fps from an average of 40fps+. They were the only two graphics cards to really drop off like this and the result was extremely stuttery gameplay.
Now at 1440p the GTX 780 was just able to stay above a minimum of 30fps, which was just 2fps better than the GTX 960. This time the R9 290 was just 1fps on average slower than the R9 390 but it was 6fps slower when comparing the minimum frame rate. The R9 390 matched the R9 290X and the R9 390X delivered exceptional performance falling just short of the GTX 980.
The GTX 770 and R9 280X struggled with less than 40fps on average, though once again it was the R9 380 and R9 285 that completely tanked, this time dropping down to just 9fps. As a result, the HD 7950 and GTX 680 provided smoother performance.
Using the maximum in-game quality settings at 4K saw VRAM usage climb to just over 6GB when using the R9 390X. Despite the massive VRAM requirements the R9 390X was just 2fps faster than the R9 290X, which is almost surprising. As we have suspected, the R9 390X simply doesn’t have enough horsepower to deal with that much data, so while it can store it, this doesn’t help the GPU get anything done faster.
For an average of 30fps the R9 290X is required while the GTX 970 just fell short, though the difference between the two shouldn’t be noticeable. The same goes for the R9 390X and GTX 980 as both are just capable to delivering playable performance. Ideally those playing at 4K will want the GTX 980 Ti at their disposal, and once the SLI bugs are worked out two cards will deliver some pretty nice results.
Republished with permission from:
Steven Walton is a writer at TechSpot. TechSpot is a computer technology publication serving PC enthusiasts, gamers and IT pros since 1998.