Gadgetory


All Cool Mind-blowing Gadgets You Love in One Place

Vulkan vs. DirectX 11 Benchmark

2016-02-18
AMD and NVIDIA have gone back and forth this week with their Vulcan announcements the API was fully ratified so version 1.0 of the Vulcan API specification has been finalized and AMD first announced its beta drivers for the Vulcan api compatible presently only with the Talos principle which is a game by Croteam and NVIDIA later about a couple hours later actually came out and said that they had their own drivers available and they very explicitly stated in the press release that their drivers were not beta unlike some other people and they were fully conformant and all these other things they did a bit of a subtle slam to Andy maybe not so subtle and we reported on that on the website and then later Nvidia came out and said we had a misprint in our press release our drivers actually our beta but they are fully conformant so make of that what you will but the point is that both companies have beta drivers available for the Talos principle and for the Vulkan API initial integration for the gaming world the Talos principle has not fully optimized its Vulcan integration the inclusion of the API is presently in very early stages and the Talos principle team has indicated that even DirectX 9 performance used to outrank DirectX 11 performance within the game but later optimizations resolved and switched over that performance so DirectX 11 now as it should outperforms dx9 the team expects the same thing to happen with Vulcan noting in its own FAQ that performance is low currently due to the early state of the build but they do expect a switchover point at which time Volken will overtake DirectX 11 for performance test methodologies in the article linked to the description below we used AMD drivers 16 dot 15 dot 1009 bida 4 and Nvidia's 350 6.39 vulcan beta drivers now these Nvidia drivers are actually technically a bit older than their most recent driver but they're the only ones that initially supported Vulcan we were not able to collect our usual 1% low and 0.1% mobile metrics because our logging software does not pass through Vulcan currently and so we were forced to use the games FPS logging system and said fortunately though the Talos principle actually has an exceptionally well built benchmark and does poor low frame rates and averages and all these other metrics that we can use in a very useful way we ran the test currently in some pretty limited scenarios so we're only looking at the 390x and 980ti and very clearly I want to say that this is not a 390x reverse 980ti benchmark first of all it's not a linear comparison and secondly where we are looking at DirectX 11 versus Vulcan not the x12 and not 980 versus 390 it's just Volk inverse dx11 on each of these cards individually and then also we looked at this on only a 59 30 K which is a good CPU to make sure we're GPU bad for the most part but one of the things with these new api is is that they should enable lower m cpus like the i threes to better perform in a way that removes some of the draw calls geometric complexities from the cpu and pushes it on to the GPU and that will improve the performance of low-end CPUs by moving their tasks to parallel processors ie GPUs so that's not something we tested just yet we want to wait for Vulcan and the Croteam game to update a bit more first and mature but for now this is what we've got and let's just look at the benchmarks right now at 1080p Ultra we observed a performance disparity of fifty seven point eight nine percent for the r9 390x favoring d3d eleven seventy five point five FPS for Vulcan and one hundred thirteen point three fps averaged over three passes for DirectX 11 and that amounts to the averages with the low performance at forty one point six FPS versus eighty eight point one FPS the nine eighty I had a performance delta of forty four point four seven that favored DirectX 11 and the nine eighty I at 1080p ultra ran 90.7 FPS average with a fifty seven point seven fps low on Vulcan and it had one twenty seven point two FPS average versus 98.3 fps low on d3 D 11 so again a favorable gain for D 3 D 11 over Vulcan moving on to 4k at Ultra settings AMD's r9 390x shows a performance disparity of forty seven point eight percent between Vulcan and D 3 d eleven favoring microsoft's established api d 3 d eleven or DirectX 11 the low performance is fairly tightly times the averages at 22.1 fps low and 29.9 FPS average versus 39.2 FPS low and forty eight point seven FPS average depending on which api you're looking at Nvidia interestingly fronts a thirty one point six seven percent performance disparity between Vulcan and d3d eleven favoring DirectX 11 and that's a significant lead over AMD and FPS Delta's at 4k sorry Kiryas to see whether Nvidia can maintain its present advantage in the deltas anyway between Vulcan and d3d 11 and again not between the nighty I of the 390x is just between the eight P is as that's what we care about so it'll be interesting to see if they can continue that lead as the API matures and as the crow team development group continues advancing its own games support of Vulcan so what does this mean for gamers well right now there's no real reason to use the Vulcan API with the Talos principle as that is the only game that supports it on Windows anyway so with the Vulkan API currently you have to opt into a beta with the Talos principle it crashes a whole lot and it doesn't really perform nearly as well as DirectX 11 and we were seeing 50% disparities in some cases so it's not something I'd recommend for anyone other than someone who's benchmarking or is an enthusiast really curious to see what Volcom looks like or maybe your developer and you just kind of want to get into a debug mode or something and check it out from another team so those are the only real reasons to run Vulcan right now but as this thing matures it could actually become a better option than DirectX 11 we'll see how that goes but in theory that's how it should work it just really depends on the driver teams at Nvidia AMD and the CRO team development group at the making the Talos principle so that's kind of what we're waiting on right now Volken and DirectX 12 do not currently exist in the same game together so we can't do a dx12 versus vulcan test right now in a linear fashion though in game DX 12 really there's a couple there's the their new fable game there's the upcoming Gears of War games should be DirectX 12 sortie out for consoles and will be coming to PC eventually I suppose but the Vulcan doesn't exist in either of those of course ashes of singularity is another big dx12 benchmarking am also does not have Vulcans so until a point at which they exist in the same title we can't do a head-to-head but for now it looks like DirectX 11 is a bit advantaged over Vulcan in a tiles principle and we're excited to benchmark the two going forward as both drivers and the game update so that is all for this video check out the link of the description below for more information on test methodology hit that patreon line if you wanna help us out directly and subscribe if you like the content I'll see you all next time
We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.