Gadgetory

we have really cool data for this benchmark it was actually pretty exciting to work with by bridging our pre-launch beta benchmarks with identical testing methods for the launch version of destiny - including some driver changes we're able to plot historical performance between Andy and NVIDIA hardware and also able to better pinpoint CPU uplift game code uplift and GPU driver uplift matching today's data against our beta optimization guides we're also able to see what graphic settings have caused the biggest gains in performance all the way down to vendor specific graphics setting optimizations before we get into that this content is brought to you by the Thermaltake flow RGB closed-loop liquid cooler which is a 360 millimetre radiator plus 3 120 fans that are RGB illuminated if then we'll take it ring fans at that this is a 4.5 done a detect pump which is one of the faster pumps you can learn more at the link in the description below there's a lot more analysis in this game GP benchmark than normally because we have so much more data to work with having the beta of destiny 2 as a baseline to work from and it gives us a way to plot scalability of beta versus launch performance in destiny 2 which includes destiny 2 game changes and of course the driver game driver changes for game ready drivers so that'll be interesting to look at and give us a better understanding of why some of the performance uplift happens the way it does because if you just look at them one after the other it can be a little confusing where AMD will lead in some settings for some benchmarks and then video will lead in other benchmarks with different settings so this allows us to pinpoint that and just to quickly recap the previous tests we did a lot of stuff with destiny 2 we had a texture comparison guide which is still relevant now with launch we had a game optimization mini guide at the bottom of the GPU benchmark that plotted the performance uplet's from different settings and basically showed that depth of field was nuking performance on everything especially Andy and then we also had some information on CP performance so we had a CPU benchmark which we have just redone for the launch version as well so you'll find that on this channel soon as for testing the difference now largely is anti-aliasing so we always tested with fxaa for the most part which made it really easy to work with and that's because MSAA is now gone basically Bungie said that MSAA was buggy in the beta and with the launch version they said forget it we're just gonna remove it from the cam that will fix the problems so that's gone we're still using the fxaa here--if which we use just originally because that was kind of a suspect move from Bungie and other than that we'll just go through the charts and you can get all the data there as always check the link of the description below for all the testing information if you're curious about the test platform used here and also one other note although averages are comparable between the beta and the launch tests you can't directly compare the low numbers that we're using because we used a different tool for benchmarking because destiny is kind of a moving target it just blocks a lot of performance analysis tools and measurement tools so we had to change once again which means that the low numbers are not fully comparable starting with 1080p highest as that's going to be the most popular you configuration for the game this configuration has us of bumping into the frame rate wall for the 1080 TI which encounters the 200 FPS captor in some parts of the benchmark despite encountering an artificial cap that 1080 TI leads the 1080 by approximately 20% that will look toward 1440p it's a better understand performance with an uncapped ceiling the GTX 1080 operates a 156 FPS average leading the Vega 64 card by 7.2 percent and marking substantial uplift from Vegas previous 122 FPS average in the beta version and with the previous and the driver revision that's a roughly a 19% performance increase from beta to full launch for Vega 64 braining 64 pass to the 1070 and now closer to the 1080 Vega 56 runs roughly tied with a 1070 SC and all walks of performance including frame time performance with a 1066 gigabyte and rx 588 gigabyte cards also near one another in the three reported metrics that we have here here's a frame time plot of the Vega 56 and gtx 1070 cards with these settings just to give a better understanding of performance as always we're looking for consistency here and note of course that lower is better but more consistent is even better than just being a lower frame time this next chart is pretty cool though this is the scalability versus beta and the line plot shows average FPS of each GPU versus the beta average FPS which was tested with earlier driver and game versions and video primarily benefits in the 1080 TI with the 1060 1070 and 1080 posting smaller gains in the range of three to six percent as you'll learn in a moment these games are primarily due to a cv bottleneck loosening and not necessarily game changes Andy's at Vega 64 and Vega 56 clearly benefited from either the graphics options changes and the drivers or both registering 12 to 19 percent performance uplift with these settings the Polaris 580 also gains reasonably the difference is enough to sway NAT Vega 56 tied with the 1070 as opposed to the previous deficit against the 1070 at 1440p highest is where the numbers start to come together he plotted a similar performance stack but note that this will switch up once we get to the high results the 1080i manages 111 FPS average posting a 33 percent lead over the 1080 FTW with Vega 64 leading the 1080 by 15% this is a little different from the previous results and Vega 56 is also functionally tied with the performance of the 1080 FTW at 1440p highest plotting an 83 FPS average and keeping low is consistent those places Vega 56 in a seriously advantaged position versus the 1080 and 1070 and if you can find it for MSRP it makes it a better price competitor for these particular settings but it gets more complex than that as you'll see in a moment because it this largely hinges on what settings are used Vega 56 does not always have this position but it does in this test as for scalability versus the beta here's what we have the Nvidia cards show almost no scaling here even less than the previous one partly because Nvidia already had game ready drivers at the time of beta launch so there's really not much ground for gtx devices to game and video already put that work in earlier whereas AMD just now put out their most ready drivers for the game they get 56 host a gain of a 50.9% an average FPS for our tests close to an these boasted 42% gains with the driver launch notes Vegas 64 post 49.5% scaling although these aren't unprecedented gains from a driver update and game update the two compounded together make for a large leap in the line plot here's where the data gets interesting if you remember the 1080p scalability charge the Nvidia cards there were posting more noticeable gains than in the 1440p scalability chart where we saw nearly zero gains and some of the tests if we bring the 1080p plot back on the screen you'll see that the 1080 Ti as noted a moment ago had the largest gains of all the Nvidia cards with the gtx 1060 posting nearly zero uplift that's because of a cpu uplift and performance with destiny to which we also have plotted in our forthcoming destiny to CPU benchmark the cpu uplift has alleviated some of the bottlenecking on the 1080 Ti hence the bigger gains but obviously hasn't helped at the 1060 level because we were already limited by the GPU this is an instance where at 1080p highest the gains look like they do on the GPU is not because of GPU uplift but because of alleviated constraints in the CPU pipe that's why the two plots are so different from one another and then of course and these gains are largely because of that driver update where the earlier version of the driver just wasn't built for destiny properly reducing settings to high but staying at 1440p we see the gtx 1070 catch back up to the RX vega 56 leading marginally with a 1080 leading Vega 64 card by 14% the 1066 gigabyte and rx 588 gigabyte cards are within a few FPS of one another with a 1066 gigabyte leading by 2.7 percent and here's that beta versus launch scalability charge it now becomes clear that Andy had the most to gain when running highest settings between launch and beta which is something that we highlighted in our original destiny to beta research and coverage again the biggest gains were specifically for AMD running highest settings with less concern regarding high during the beta we route caused this to depth of field on highest which tanked and the GPU performance specifically but was also a big hit to Nvidia you can see that in our beta graphics optimization guide from the original launch of the beta version of the game changing only depth of field from highest to high could get us 82% of the way to switching all options to high this means that you could leave everything on highest from the highest preset and change only depth of field and get 82 percent of the performance of just dropping all of them to high that means that something in an deist drivers or destinies code has resolved this problem for Andy Hardware specifically and that's the explanation for the performance we're seeing our final few tests are at 4k with the highest plotting the 1080 Ti about 1 to 2 FPS ahead of the beta at 56.8 FPS average and that leads the 1080 by 36% and the Vega 64 by 12% Vega 64 out matches the 1080 in this instance once again and that's basically coming down to the settings in terms of scaling Vega 64 benefits the most moving from 32 fps to 50 point 6 FPS average with the launch drivers and launch version of the game pretty momentous changes for Andy's Vega cards at 4k high as the previous tests have informed us this changes again Vegas 64 is now behind the 1080 with a 1080 leading by 13% the 1070 is now nearly tied with Vega 56 with Vega 56 slightly ahead this was all conducted in what is functionally a single-player part of the game or at least a co-op version it's the campaign more or less not PvP but we did do some multiplayer as a single-player testing in the beta and found that for the most part they were fairly comparable in terms of performance particularly in terms of performance scaling which is the part you should care more about than just the raw FPS number what you're looking at in these benchmarks is what kind of scaling do I see from one GPU to the next not necessarily what hard FPS do I see from one to the next because obviously that's going to change depending on where you benchmark in the game and this game has a pretty wide range of framerate depending on where you're testing for the most part it is fairly consistent when testing for example the first 5 minutes of the game in the beta and now with the 1080 TI we saw at 4k highest settings in the range of 58 FPS average both times beta and launch version of the game and that was consistent throughout that whole period however every now and then it's not too often but every now and then there's a scene that is either exceptionally high-end performance or sort of low-end performance not too much of the exceptional low-end and that means that depending on where you're testing obviously it'll change a bit but the scaling seems to be pretty much linear between all the devices so that's what you want to look for the most for multiplayer and singleplayer testing by which I mean pvp competitive versus the rest of the game we did see generally comparable performance and you can check the previous beta benchmark for that but we we're probably not gonna redo all of those again because the scaling is pretty much the way it was before with the driver and game changes counted as well so it's pretty much the same performance between different net areas of the game the scaling you see is from the updates though and that applies everywhere globally to the game so there's some cpu uplift here as we plotted at 1080p highest and the depth of field had major GPU uplift what we're not sure of if is if the other options in the game between highest and high are impacting one GPU vendor versus the other more heavily but depth of field definitely has had a pretty big change since the beta because you can plot that trend and that's why Andy's getting a lot of the uplift they're getting in some of the benchmarks particularly the ones with highest settings because depth of field has that difference and Vidya had minor uplift and it's more likely attributable to game code changes may be some minor driver tweaks although that seems there's not a lot of that if there are any because the original drivers were pretty much out of roughly a hundred percent of the performance of where this is today the exception being 1080p where we were becoming CPU constrained and that has now uplifted a bit with the launch version of the game so that's all for this one as always you can get a patreon.com slash gamers access to support this type of testing if you like it subscribe for more will have a CV benchmark as well and check out story rock gamers and access that for a shirt like this one I'll see you all next time