Bethesda’s recent Doom reboot has been a huge hit, even despite significant misgivings about the game’s multiplayer mode. The single-player game is a fast-paced return to Mars (and Hell, natch) that plays beautifully and fluidly. Now, the already-polished game is getting faster, courtesy of a major Vulkan patch. This makes Doom the first game to prominently adopt Vulkan as a way to further increase performance rather than simply as a proof-of-concept implementation.
According to the Bethesda forum post:
We also anticipate some older GPUs will now be able to play the game at good frame rates. We hope the range of GPU support widens with additional game and driver updates. That said, this is the first time a triple-A game is releasing on a brand-new API and brand-new drivers, so there may be a few bumps, but our testing is showing really great performance and stability.
There are, however, still some caveats to be aware of. Bethesda notes that the Vulkan implementation currently doesn’t support any 2GB GPUs from Nvidia while running Windows 7 (there’s no mention of a similar restriction on 2GB GPUs from AMD). The GTX 690 is also unsupported. It’s not clear if this Nvidia card restriction only applies to Windows 7 or if 2GB NV cards also run into trouble on Windows 10. There’s more information at the website FAQ.
Prior to this patch, Doom already ran extremely well on both AMD and Nvidia hardware. Earlier coverage by TechSpot (pre-Pascal) showed Nvidia with a strong 1080p advantage that shrank as resolutions rose. The GTX 980 beat out Fury X at 1080p (122 FPS versus 118) but was soundly beaten by that GPU by 4K (47 FPS vs 38 FPS). Nvidia has been silent about any performance improvements in Vulkan, but AMD has published a blog post with additional performance data and information on the API implementation.
According to AMD, Doom now takes advantage of asynchronous compute and shader intrinsics that allow the game engine to directly access hardware (a video describing how this works is embedded above). It’s not surprising to see AMD picking up significant gains in Vulkan; the new API is a descendent from AMD’s Mantle and functionally equivalent to DirectX 12.
According to AMD, the RX 480 is up to 27% faster in 1080p when using Vulkan and 23% faster in 1440p. The fine print attached to those figures states:
Testing conducted by AMD Performance Labs as of July 6th, 2016 on the AMD Radeon RX 480, on a test system comprising Intel i7 5960X CPU, 16GB DDR4-2666 MHz system memory, Radeon Software Crimson Edition driver 16.7.1 and Windows 10 x64 using the game DOOM on the ultra preset. PC manufacturers may vary configurations, yielding different results. At 1920×1080, Radeon Software Crimson Edition 16.7.1 running DOOM OpenGL scored 106.40 and Radeon Software Crimson Edition 16.7.1 running DOOM Vulkan scored 135.65 on AMD Radeon RX 480, which is 27.5% faster performance…. At 2560×1440, Radeon Software Crimson Edition 16.7.1 running DOOM OpenGL scored 68.51 and Radeon Software Crimson Edition 16.7.1 running DOOM Vulkan scored 84.34 on AMD Radeon RX 480, which is 23.1% faster performance.
The RX 480 is just one GPU, and we’ve already discussed how different cards can see very different levels of performance improvement depending on the game in question — the R9 Nano picks up 12% additional performance from enabling versus disabling async compute in Ashes of the Singularity, whereas the RX 480 only sees a 3% performance uplift from the same feature. Nvidia has not released a statement indicating whether or not the company expects to see a performance uplift from using Vulkan or which GPUs it expects to see the most benefit but testing by Guru3D suggests that the 1070 is completely flat in Vulkan vs. OpenGL. Guru3D’s testing also shows that the RX 480’s performance improvements are in-line with AMD’s claims.
One of the most significant differences between Nvidia and AMD over the past 12 months has been their approach to next-generation APIs. AMD has prominently talked about DirectX 12 and Vulkan at every opportunity. That’s partly because both APIs are closely related to AMD’s Mantle, which debuted a bit less than three years’ ago, but it’s also because DX12 has often improved AMD’s GPU performance relative to Nvidia. Team Green, in contrast, has had comparatively little to say about the API shift. Pascal’s vastly improved preemption capabilities obviated the performance hit that Maxwell took in some asynchronous compute scenarios, but Nvidia isn’t really putting much of a PR push behind the API shift. Visit the DirectX 12 section on GeForce.com and there’s just one title listed there, and nothing about games like Hitman, Rise of the Tomb Raider, Ashes of the Singularity, or Doom itself.
We’re not implying that Nvidia isn’t committed to DirectX 12, but the company’s public messaging on low-overhead APIs has been anemic these past 11 months. Pascal’s excellent performance has given Nvidia the overall lead at the top of the market, but there are millions of 28nm GPUs from Teams Red and Green still in the field. 28nm GPU performance in DX12 is going to be a topic that people care about for a long time to come, and AMD’s overall messaging on that topic has been stronger than Nvidia’s thus far.