Gadgetory

we first thought dhoom running the Vulcan demo last week but it's not out yet and Vulcan can be expected to be added to doom for its API support list probably at a couple weeks front we've been told by in software and we saw that and we're told that at the NVIDIA GTX 1080 unveil events of course we'll be reviewing that shortly so that leads to a quick disclaimer before getting into this benchmark because of the new architecture is right around the corner that BGP 10 for the GTX 1080 and 1070 and then and these Polaris architecture because those are right around the corner if you're watching this to buy a GPU for this game you may want to wait a couple of weeks and just see how the new cards perform because it's very likely that there will be a lot better than what we have now we're already working on it but that'll be online shortly so let's get into this this is the doom benchmark we tested all of our GPUs from 1080 to 4k resolution including 1440 in the middle and we ran on ultra settings with a slight splash of medium settings for the low end cards some of the gameplay footage you're looking at from our video is running at 4k ultra on a GTX 980 TI just for reference and let's start right off with the testing methodology and then move along to what the graphic settings mean as we normally do first of all the components table shown on the screen now is what we're using for our bench platform and that's the x 99 benchmark that we run for these games that's fully detailed in the article link to the description below we used level 2 for benchmarking and that is called the foundry the foundry seems to be one of the most GPU intensive areas of the game it's heavily dotted with bloom effects and directional occlusion mostly seen near the lava flow and there are a ton of decal elements toward the front of our benchmark presence from previous combat and prefab Elvins this adds some to the vram hit although that's marginal most of the heaviest effects in Doom take place after the geometry and rasterization process in the pipeline so that means it's all pretty much press processing and particle effects that are computed it later we use the latest 1652 AMD drivers and 368 146 unreleased nvidia drivers which will be using for the gtx 1080 benchmark once that review goes live testing was conducted at Ultra settings at 1080p 1440 and 4k and we used fxaa for the aliasing we did not use temporal anti-aliasing for this test but did check it briefly and saw performance at upwards of ten percent when using t SSA eight tap which is sort of a GPU agnostic version of TX AAA the temporal and video solution for anti aliasing as always we conduct a 30-second benchmark three times for parity / setting for GPU so again three tests / configuration per card some medium tests were briefly conducted to get an idea for low end performance let's get to the results at 4k Ultra doom proves abusive on the GPU and shows top-end performance of 45 FPS on the GTX 980 TI with tightly x 1 percent and 0.1 percent low metrics next the r9 fury x pushes 4167 fps and has loser frame times but is still reasonably tightly timed and the performance gap between the 980 I and the fury X is roughly seven point seven percent just for an idea to get reliable 4k performance near 60 FPS you have to run these cards at a mix of medium to ultra settings with no real anti-aliasing going on for the most part unless you want to tank some of your other settings fxaa is okay though moving the 1440p we see this generation of Maxwell Hawaii and Tonga GPUs are capable of sustaining 1440p at reasonably playable frame rates the GTX 980 TI tops the chart again and that's a reference clock by the way with 86 33 FPS average 67 dot 33 one percent lows and 60fps 0.1% low is very tightly timed there and that is seventeen percent ahead of the GTX 980 non ti which runs at 72 @ 67 FPS average and that is flanked fast by the fury X running at nineteen point five percent slower than GTX 980 TI or two-point-three percent slower than the GTX 980 non TI all cards for the GTX 970 and up are playable and able to sustain 1440p frame rates at ultra including the r9 fury x which we've just added to our bench and the r9 390 x is more or less there but would hit 60 FPS with some slight settings weeks and i do have notes on the 390 X in a moment finally at 1080p we see some interesting performance metrics crop up the GTX 980 TI pushes 1 30 FPS ish with the i7 5930 kcp were using on our bench platform and these are 9 fury x is lost it's lead by the time we get to 1080p and that is just because the fury ax and other AMD devices tend to do well with abusive pixel throughput just raw throughput of pixels through the pipe and the fury X still holds fast against 10 videos best cards below the high end cards the near 60 FPS hitters include the r9 390 x gtx 960 and r9 290x a bit old now and these are 93 90 x and 290x get obliterated by the gtx 970 though and that's a performance gap of about thirty nine percent and the 290x is even getting beaten by the gtx 960 which obviously that should really never be the case with a properly optimized game and or driver sets so we're not sure who's at fault here but it's clear that either doom is poorly optimized on the softer side or and these 290x drivers aren't working too well and we retested this several times the 290x and the 390 X because it also shows anomalous performance and validated on another identical machine but still saw the same issues so this is not a platform issue this is not a software issue on our end it's something going on with doom or they am the drivers and we didn't use the latest AMD doom ready drivers some web searching did show that users of the 390 accent 290x have experienced similar performance anomalies and frame rates which coincide with our findings so it seems like a somewhat known issue and hopefully will be addressed soon the GTX 970 the far and away outperforms r9 390 X right now and performs effectively as well as the fury x4 1080p anyway the only reason that 39 DX holds its own at higher resolutions at least against Nvidia devices is because of the raw pixel throughput that I talked about with the fury accent ago Andy tends to do well in such scenarios but that's their brute force victory with the pixel processing for 1440 and 4k resolutions and it seems that a driver update or game optimization update is in order to maximize am these abysmal performance on its through 90 x of 290x the fury x isn't exactly where it should be either but it's doing pretty well and is at least somewhat close to the 980 tied in some instances and that's not terrible for the price though certainly you would want better out of a card that's more expensive than a 980 hopefully though the drivers will resolve that just for sake of sanity here's a quick table non chart of a couple medium tests we ran just performance at the medium settings the 950 and that leads us to the next and final question which is what video card is best for doom and of course that depends on your use cases and the resolution for 4k you'll want either the 980 I or fury X you'll see that our charts show at North a 40 FPS for ultra but you'll have to drop that settings configuration from ultra to something else maybe a mix of medium high and ultra and that will allow these 60 FPS throughput that most of us desire especially for multiplayer gaming and of course our benchmarking was done in single-player so your mileage may vary with different multiplayer maps although the course was fairly intensive as far as 1440p ideally you're running something like a GTX 980 a fury X or GTX 970 and I would mention the r9 390 accent here as well because it really should do pretty well it's basically at 60 FPS but the performance is just really a lot lower than it should be for 390 X and that's some kind of mixture as i stated of the drivers or the game software in terms of optimization and that's just because the GTX 970 is obliterating the 390 X right now although the fury X isn't too terrible but 980 is better value as far as 1080p we're able to run something as lowly as the 960 and the 380 x and this is where it and these newer architecture does actually prove to pull its weight a bit better so the 380 x runs pretty well at 1080p ultra with doom and the gtx 960 runs very well at the same settings for doom so those would be sort of your bottom line if you want to sustained high FPS and basically all scenarios including the most intensive combat and geometrically complex scenarios then something towards the 970 would make sense but again as stated earlier if you're going out to buy a card sort of in the next week or so maybe next few weeks and you want to buy it for doom I would say wait for two reasons one they might release some kind of padua one of course will be coming out and that may show different performance for AMD vs nvidia and then game optimization patches may come out which would resolve this weird 3 90 x issue and finally the main thing to look forward to is just the the release of the gtx 80 which is known to be made 27th the 1017 on to be June tenth i believe and then AMD's polaris we don't have a release date on that yet but it should be entering the market at the budget to mid-range class and that will be in the next few months maximally so those depending on what you've got right now may be worth waiting for but that is all for the doom benchmark check the video in the poster all if you want to see our benchmark course again with some brief explanation of graphics as far as the more in-depth stuff really the main items are directional occlusion and that's all done in the G buffer you can read about that in our article link in the description below if you're curious about how that works and then the anti AAA scenes the other sort of major thing that people may not be familiar with so you've got the normal morphological anti-aliasing and that's SMAA and there's taa which is temporal anti-aliasing which means that it looks at the frames one to the next so frame a versus frame be the next one looks at them determines which colors are appropriate for which edges based on movement that's a semi new technology and pretty important TX AAA is not here but t SSA is here and that is another temporal form of super sampling anti-aliasing hence t SSA that is used agnostic leon AMD and NVIDIA devices and is also temporal so it's over time frame to frame those are the main things and here people may not be familiar with the only other one that stands out immediately is chromatic aberration which is exactly as it is in photography just like that's the field chromatic aberration is sort of a lens effect that you get so if you think of those Michael Bay movies where you see a lot of lens effects that would be there's some chromatic aberration there especially with red tinting on the lens depending on how it's perceived the light and again article below if you want to read more about that so thank you for watching pay challenge post your video to help us out directly i'll see you all next time