DOOM Vulkan Performance Analysis

1080P will have the largest increase, as is natural, but just how much faster is it with Vulcan enabled. We have to note that FXAA disables Async compute, so we used TSSAA throughout, which we used to test our previous results anyway.

DOOM Vulkan

DOOM Vulkan

Looking at the results we see that AMD has a significant increase in performance. Vulkan accounts for a nearly 61% increase in performance the best case scenarios for the RX 480. That is, the level we played through. The increase is more for the Fury X, which doesn’t quite have the same enhancements done to the geometry engine or each individual compute unit that Polaris brings.


If we we take a closer look an compare the performance of the Fury X with that of the GTX 1070, arguably it’s closest competitor in this situation, we see the advantage is clearly in AMD’s court. Performance is of the charts, so to speak. It almost raises the question of whether or not GCN was created as a next-generation project, made specifically with DX12 and Vulkan in mind. This kind of increase in performance is something that not all games with the API will enjoy, and it doesn’t mean that Pascal isn’t suited to Vulkan, just that it isn’t suited here, in DOOM. The lack of asynchronous compute being enabled in the past was curious, and here it seems that even with it, there’s no advantage on NVIDIA’s part. This may be a result of optimizations more than anything. Let’s up the resolution now.



Upping the resolution lowers the advantage appreciably, instead giving us an improvement of 57% for the Fury X which is still rather large, considering. The GTX 1070 receives a marginal improvement, though it’s still one of the fastest cards in the bunch. Crucially, there’s no decrease in performance for any card.



4K sees the advantage become much smaller, though that’s due to many factors. The scenes just become incredibly intense with a tremendous strain on even the RX 480 with it’s 8GB of VRAM. The GTX 1070 remains in the lead, though only just. It stands to reason that should improvements in how NVIDIA’s driver handles context switching be implemented, we should see a much better advantage with NVIDIA. Until then, however, we have what you see before you.

What can we conclude?

Vulkan was just officially released out of it’s beta stage only recently, and until now we havcen’t quite had any examples that provided anywhere near this level of performance. It’s interesting that even with async compute supposedly enabled in NVIDIA’s Pascal architecture now, there’s still only a marginal increase in performance when switching from OpenGL.

This is a very large boost for AMD and enables you to play the game with the highest visuals on even a 380X at 1080P. Even better is how the minimum framerates are appreciably increased as well.

Vulkan is still in it’s earliest of stages, and the improvements in performance are dependent on the developers themselves, though this makes it seem as if there’s hope for high performance and¬†gorgeous visuals. It also could drastically shift the value in GPUs. An RX 480 and potentially the RX 470 and 460 have just gained in value simply because they have more potential to provide a better, more playable experience than ever before.

NVIDIA just needs to work on how it handles context switching and how it handles the various new technologies in Vulkan. Technically they should be driver agnostic, though that isn’t always the case when the software is acting as a sort of intermediary in this case. Further driver improvements will see more performance increases for both parties, NVIDIA specifically. We look forward to that. Until then, we have what you see here. Which is quite nice indeed.