Gadgetory

in today's battlefield one benchmark we're looking at API performance between DirectX 11 DirectX 12 on GPUs from both vendors some of which are not shown here and part of this look is to determine whether or not DX 12 actually improves performance on the whole with battlefield 1 because as you all know it's not always a positive gain before getting to that this coverage is brought to you by an tech and a new cube many ITX case as designed by Razer and it's a small box that supports a full custom loop liquid cooling solutions in today's battlefield 1 benchmark we're looking at GP performance between the GTX 10 series rx 400 series GPUs some of the 300 series GPUs and we've also got a couple of 9 and 7 Series TVs one of them anyway the 750 Ti this game as with all EA origin games does have an activation limit when you're dealing with Hardware swaps for 24 hour periods 5 cards per game code we have three game codes to maximum of 15 cards we can test assuming no screw-up so we are more limited on what we can do because of the EA origin de nouveaux stuff let's talk methodology first since this is always a question this will be more intricately detailed than the article linked below if you're curious about why or how we did something check there first because it probably answers the question partner cards were used where available and tested at for out of box performance frequencies listed are advertised clock rates we tested both dx11 and dx12 using present Mon by Intel Microsoft for the DX 12 testing and we execute present mod using command line then use our own in-house Python script to rip out the average the 1% loan the point 1 percent low FPS which illustrates stutters as we've explained before please note that we use on present to measure frame times and frame rate and methodology of course is important so make sure you understand what you're comparing results to if it's odd present on display or even measured using the same tools at all we use on present and present Mon for DX for benchmarks also note that we are limited again on the activations for game code so keep that in mind that battlefield 1 has a few critical settings that require tuning for adequate benchmarking except where otherwise noted we disabled the GPU memory restriction for testing because that triggers dynamic quality scaling creates unequal tests and we set resolution render scale to 100% fields of view was also changed to 80 degrees to more appropriately fit what we think a player would generally use on average and that's from a default of 55 so this impacts FPS of course and that's also important vsync an adaptive sync are disabled presets are used for quality as defined by the chart titles game performance also swings based on test locations that matters to if you're comparing generally a good idea by the way not to compare benchmarks between multiple testers without really understanding where and how the things are tested we tested in the Italian avanti savoia campaign level sorry for the pronunciation that's single-player and we also tested on Argonne forest for multiplayer you can view our test course in a separate video already on the channel the campaign was used as a primary test platform but we tested multiplayer to determine the scaling between single-player and multiplayer thankfully two are pretty comparable in performance depends heavily on the map as always but even 64 player account servers assuming the map is arranged as usual where you never see everyone I want it's not too abusive on the GPU here's some of that preliminary data the game actually scales fairly linearly between our test cases single-player and multiplayer and note that as always specific game events or the introduction of a lot of geometric complexity at once like multiple tanks on screen at a high LOD will impact FPS similarly flying up in the vacant sky or staring at low-density objects will impact FPS in the opposite direction for our tests we have a planned mix of geometry effects crepuscular rays and in-game events the chart we've been showing reveals the scaling between our multiplayer non-repeated real gameplay benchmark and the repeated single-player gameplay scenario raw performance looks like this table on the screen now if you want to see a couple of raw numbers and that sort of spoils some of our performance results so let's just jump right into DirectX 11 with 4k at high settings we briefly tested with 4k at Ultra settings as well to compare versus our final use of high settings with 4k the two are actually pretty similar in this scales to 1440 1080 as well generally only producing a couple of percentage points difference a couple frames different but still we settled on this chart for 4k high testing EVGA is GT X 1080 FTW hybrids to the top of the bench pushing an average FPS of 69 with 1% low 63.1% loads of 56.7 and that's an indicator of consistent frame times with low latency and variance the gtx 1070 SC operates at an average of 54 FPS for 4k high with lows dip into 44 fps the fury X is not far behind on these charts and following these the msi r9 390x and gtx 1060 gaming x cards are effectively tied in their averages and 1% lows with the r9 390x imperceptibly favored in the 0.1% blows rx 488 gigabyte gaming X is operating at 37 FPS average at 4k high and these three cards the 1060 390 X and 480 are bordering on unplayable for a higher speed shooter like this one if you really wanted 4k gaming on these devices which is probably not common you would want to dip the settings closer to medium let's look now at how the same settings perform under dx12 with DirectX 12 enabled and using the same settings otherwise the GTX 1080 is now operating at 59.6 FPS average or about 10 FPS lower than TX 11 and was significantly worse 1% in 0.1 % lower values in DX 11 this is precisely why we use this testing methodology because those stutters would not appear in the data if we just used averages you'd be relying on subjective analysis and the stutters by the way our perceptible and gameplay a sudden hitches and frame latency you'll feel a delay frame the frame occasionally in the test passes but not always and in this instance DirectX 11 should provide a smoother performance or experience overall looking at the next card down the gtx 1070 sits at about 48 FPS average with similarly low 0.1% performance trailed next by the r9 fury X and r9 390x the fury X has the worst frame time performance on this chart we suspect that this is at least partially a result of its more limited vram both 390x is really not much better and we believe that some of these frame time hits are a result of dx12 memory management falling more on the is of ease then on the IH vs or the difference of basically software vendor versus a hardware vendor the ISPs are still learning how to deal with memory allocation for games with DX 12 Vulcan and other low-level api's so there's likely some performance and optimization issues in there to be worked out but we don't know the full extent of how battlefield was built on dx12 let's move back to DirectX 11 with 1440p and ultra settings this time not high the GT X 1080 FTW hybrid is chart-topping with nearly 120 FPS for those of you with 120 Hertz displays if that's relevant it's running 1% in 0.1% low values north of 60fps and has followed most immediately by the gtx 1070 SC at 91 at seven FPS average also with tightly time to lows really almost everything we're showing on this chart is fairly playable at 1440p with ultra settings including the RX for 80 gaming x the RX for 70 isn't really the best option but you could I guess tune settings down too high to hit that 60 FPS marker if it's really desired and granted again most rx 470 owners probably are not running 1440p displays in addition to their mid to low end card I guess mid-range would be more appropriate anyway the r9 fury X outpaces the 1060 by about 8 FPS or about 12% with the r9 390x following the gtx 1060 by a few frames the cards performing in these 60's will occasionally dip when major in-game events happen but should generally remain playable at 1440p bouncing back to DX 12 at 1440p ultra we now see the GTX 1080 FTW hybrid is performed about 22% worse and it's averages than with dx11 when we tested the same card in the same settings in the same location the lows also take a big hit dipping down to 22 FPS and the worst few frames of the benchmark not the minimum but the worst selection of the 0.1% lowest frames and this is true across the board with Nvidia holding slightly higher low performance than Andy but both are disagreeable so really it's irrelevant who is higher than whom because you're dealing with just poor performance in the lows across the board the x12 performance in this case is not beneficial to either vendor at least not for the early stages of level we found that once you kind of get into gameplay and multiplayer things do smooth over and performance improves a little bit but you will have to endure a couple of hitches in the first couple of minutes the x12 performance remains poor all the way down the stack when looking at beyond present variants and then on display would be a bit different but not too much not every single test run exhibits the variance but it happens minimally four out of five times the rx 40 for instance had one pass pass with 1% values at 61.5 FPS in 0.1 at 69 FPS both completely agreeable though the other test passes each exhibited at least one or two noticeable stutters outputting six FPS and the point 1% blows in about 30 and the 1% low is 1080p is still the market Dominator with more than 80 percent of the market still on 1920 by 1080 displays as we understand it with DirectX 11 and ultra settings 1080p resolution debuts larger charts for us that contain everything down to a 750 Ti and an Rx 460 and the GT X 1080 FTW hybrid is beginning to approach the game's 200 FPS frame recap before console modifications trailed by the gtx 1070 at 125 FPS average then the r9 fury x at 100 FPS average all devices on the bench have tight frame x from both AMD and NVIDIA other than of course the really low-end cards and as for those at lower end cards you're looking at basically a 750 Ti sub or around 30 fps you'd want to drop to maybe medium settings and the RX 460 and the 40s or so so again probably drop to medium or maybe mix of medium high settings up and what you're going for and if it's a 2 gig or 4 gig our rx 460 as for cards closer to a baseline 60 FPS for 1080p ultra you'd be able to get away with an Rx 470 pretty easily with its 75 FPS average or a GTX 960 if you've got one already right around 60 FPS average here's the 1080p dx12 benchmark chart you're looking at more of the same where DX 11 is outperforming in the one 47.1% low values and depending on which men are you're looking at maybe even in the averages generally we are recommending DX 11 right now for battlefield 1 although again if you play for a bit it will smooth over at least slightly and as for one final chart this is a scaling chart showing the performance gain or loss from low medium high and ultra this is Amalgamated from a couple of cars but we didn't test on all of them so it just gives you a way to extrapolate performance if you're trying to run it to lower setting for a certain card as for the best cards for battlefield 1 we're happy to report that with this game with the drivers that are out by the way all game ready drivers the frame time variants is really not a major issue with DirectX 11 for AMD or for Nvidia so both of them are pretty similar in that regard and DirectX 12 is really the only place where it becomes a variable but the low performance is really technically it's worse on Andy but they're both so bad that it just doesn't matter some apps are less bad than others sometimes it gets smoothed out etc but because averages are effectively identical for AMD between the x12 and D X 11 and because they're better for NVIDIA with the X 11 and as a third point because DX 12 has more variants will be basing this conclusion on dx11 performance again even without the variance the average framerate is effectively equal on AMD between the two API is in a slightly negatively scaled on nvidia this will make the performance and dollar arguments a little bit less complicated for 1080p with ultra settings owners of last gens at gtx 960 and r9 380 x will be able to play reasonably well with a few small tweaks down to high of 60 FPS is strongly desired the RX 480 and gtx 10-6 these are both more than capable performing at 1080p with the RS 480 a few frames ahead in DX 12 though with worse stutters and the gtx 1060 ahead in the x11 both of these $250 cards scale well in 1440p and are fully capable of 1440p at Ultra settings where we see frame rates north of 60fps for each vendor if you wanted to go cheaper for 1080p than the RX 470 or RX for 84 gigabyte cards would both be capable of handling 1080 ultra with FPS greater than 70 FPS average and videos GT x 1050 and 1050 Ti are also worth considering but we can't post through his reviews or benchmarks just yet aside from the DX 12 variants the game is fairly well optimized for either vendor with again dx11 it's really not bad on either vendor you could use a gtx 970 or up for 4k high or an r9 fury accent up for 4k high if you've already got one of those and then we'd suggest looking at your ideal card's performance and other games obviously if you're buying a new one today for this game check out performs with other games to make your decision because for this game really they all do pretty well and they're really not too far apart from each other but against recap 1080p is easily done with an RX 470 gtx 960 one could speculate that a 1050 TI would probably be okay on 1080p but we don't know just yet and then 1440p fine on a 1064 84 K 10 so DP reacts and that's really all for this benchmark so as always patreon link the post roll video to help us out directly links the description below for more information test methodology drivers used why we use the different cards that we did clock rates test location present Mon all stuff that's always down there is in the article so you've got questions check they're probably answered thank you for watching I'll see you all next time you