Gadgetory

the arrival of destiny two's beta on PC provides a new benchmarking opportunity for GPUs and CPUs and will allow us to plot performance uplift once the final game ships aside from being a popular beta we also want to know if Bungie AMD and NVIDIA work to further improve performance in the final stretch of time prior to the official October 24th release date for now we're conducting an exploratory look at multiplayer versus single-player test patterns and results for d2 quality settings and multiple resolution testing before that we partnered with LastPass to bring you this video LastPass is a password manager that helps generate random passwords for each account insuring unique passwords for every log in as fast strengthens account security so that you're no longer using the same password for multiple sites learn more at the link below we front-loaded this content with research done on destiny 2 to determine the best test patterns that are repeatable and reliable and easy to execute and all that so part of this was conducting tests and multiplayer and singleplayer in the beginning of the game where we wanted to benchmark in single-player later in the game and ran tests across the entire 20 minute or so tutorial campaign that ships with to determine how performance scales later to make sure there's not just some early on really heavy performance hit or really light performance hit in the game and then also how does that compare to multiplayer to make sure we're looking at a realistic scenario so we've looked at all of that a couple of things here first of all this is a beta everything is subject to change they have about a month or so until the game finally ships maybe a month and a half that means that things can change MSAA in particular as noted by Bungie themselves is still under active development and is a work in progress so we've actually not used it for this testing for that reason but that just it goes to show that this game is under development everything's beta these results pertain to beta and the beta performance on the driver side and video has game ret are ready drivers out so we use those and D has 17.8 2 which is what we used we asked them what to use and that's what they recommended and and the may push more drivers later probably closer to the official launch of the game at which point we'll revisit it and be able to plot the performance uplift from the beta to the final version and then hopefully find so many any visual differences that may exist as well so that's what this is for this allows us to see how the game improves how the drivers improve things like that and in the very least satisfy an immediate interest in how the beta is performing on different Hardware we're starting with GPS today we have CPU benchmarks as well there as well go live not too long after this video so subscribe for that if you're interested but let's get straight into the research part of this before we guess the comparative benchmarks first of all the big question was how well multiplayer and the reality of framerate and multiplayer is reflected by single-player if we wanted to run single-player benchmarks because the more repeatable and easier to do without having to wait for matches and maps and also how this compares to a session over say 20 minutes or even five minutes we've got a log that long as well so this was done with a 1080i sc2 at 4k and highest settings at first we did this across a spread of multiple settings and cards and we'll look at comparative benchmarks next where we set all the same configurations for everything's that is probably comparative but first we wanted to find what works for a benchmark this first table shows each spot measurement against one another then again the final benchmarking course overall our FPS range is consistent there's one lower FPS number out of all of these which occurred in a densely packed city street area with complex geometry we only saw this FPS dip at one time during the campaign and it seemed tied to that specific zone when there was heavy combat and a geometrically complex area averaging the spot checks across the entire 20-minute session we end up at 51 FPS average 45 1% lows and 41 fps 0.1% lows which represent our frame times in a converted FPS metric comparatively our chosen benchmark course selected forged accurate repetition ended up at 55 FPS average 48 1% lows and 47.5 0.1% lows these numbers are remarkably close and show that in the very least our benchmarking method is representative of the entirety of the beta intro campaign at these settings with this card let's look at how this compares to competitive multiplayer matches taking spot measurement across several multiplayer sessions where and we actually tried to compete and play the game properly on the Midtown map we end up with this new table of results let's put this next to the single-player results for comparison the range is 53 to 57 FPS average and averaging all the spot-checks puts us at 50 4.5 with one percent lows at 46 and 43 FPS 0.1% lows these are close enough that single player benchmarking looks accurate to multi player performance in the beta state of the game with this particular configuration to get a broader sense of scalability across the games various maps and game modes here's an msi gtx 1050 TI OC edition at 1080p and high settings our benchmark course outputs a frame rate of 70 FPS average with lows at 50 4.7 1% and 53.7 0 1 % compared to a pair of multiplayer matches on mid-town once again we see frame rates of 74 average 63 1 % and 59 0.1 % the difference is approximately 6% considering the trade-off is significantly more reproducible results that's really not too bad you'd see a bigger range of variants in a game like Battlefield with 64 players although it's possible that destiny 2 is final iteration could show greater swings across other maps and game modes will have to reserve judgment on that aspect until it fully shifts and here's the last round of exploratory tests conducted on the rx 580 which were run as we were finalizing its comparative results shown momentarily at 1080p and with highest settings our benchmark output had us at 85 FPS average 66 fps 1 root sound lows and 62 fps 0.1% lows as a reminder we're using FXAA here multiplayer testing put our range between 67 and 88 FPS which is huge and largely demonstrates why multiplayer testing is difficult to do accurately and averages out to 79 FPS average 66 FPS to 1% lows and 62 fps 0.1% lows still reasonably close with the final benchmark output averaging a bit higher than multiplayer though there's so much range in multiplayer results that we'd have to run another dozen or two times to get a better idea of the average and now that we've gone through this we know that the game is remarkably consistent in its performance across single-player and multiplayer and across multiple levels of single-player and even multiple test durations we tested at a intervals of 1 minute 5 minutes 20 minutes and so on with breaks when they were loading screens in between and saw that the averages were pretty much the same aside from that one zone which was really geometrically complex but we ended up not testing therefore the final benchmark because while it was in a weird spot of the game and - it didn't match up with the rest of the testing so the next thing to note before the comparative section there's a 200 FPS cap on destiny - so it is not unlimited or unrestrained in the purest sense of the word but basically is and this 200 FPS cap means that as we approach the upper limit of framerate the cards that exceed 200 FPS will end up locked to haunt 200 FPS and therefore will produce lower averages as a results so if you could exceed 200 obviously the averages would be dragged up word so keep that in mind that this mostly well actually only affects the 1080i for the most part but it's worth noting as always we can also only speak to the results that we've encountered discovered tested here and this is a beta game testing leaves a lot of room for different results if you change the settings we have a part at the end of this video talking about a few settings in particular that were significantly impacting results so depending on if people turn those on or off you'll see major swings in performance and that of course matters just as much as the hardware does getting it's a comparative benchmarks between the GPUs now we open with 1080p at the highest settings with FXAA bypassing MSAA as it has known issues in bungees bug log that's a work in progress with msaa destiny to beta proves playable on most Hardware at 1080p even with the highest settings enabled the 1080i sc2 chart tops here unsurprisingly with a 168 FPS average and frame times consistently represented in 123 fps 1 % and 94 FPS 0.1% we're bumping into the 200 FPS cap here so that is going to be a bit lower than if there were no cap Vegas 64 stock lands at 123 FPS average trailing the GTX 1080 FTW by about 17% and effectively tying the GTX 970 SC note again that Andy is at driver situation may yet mature for this game so we can only speak to these results as of the beta with the 17 out of 8.2 drivers will revisit closer to full launch as of now Vegas 64 ties the 1070 and our Vegas 56 hybrid mod isn't far behind at about a 6% deficit to the Vegas 64 reference card the MSI GTX 1066 gigabyte card operated a 93 FPS average planting ahead of the 970 SSC and a fair bit behind the 1070 and rx Vega 56 the many GTX 970 owners can rest assured knowing that their card is still plenty usable for the highest settings at 1080p which makes sense as Bundy did officially recommend the card in the recommended specs listing rx 580 performs similarly to the GTX 970 which means that overclocked our X 480 owners can expect comparable results to the rx 580 and the destiny 2 beta just a bit behind for stock clocks the RX 570 follows Andy's trend of showing clock speeds can account for seed deficits as the 570s performance would pull close to the 580 s performance when overclocked finally the GTX 1050 Ti and rx 560 cards would benefit from a mixture of high and highest settings if at 1080p II as they begin to drag down below the 60 FPS threshold speaking of dropping settings here's 1080p with high settings the RX 560 improves by about 10 FPS or 26 percent and the MSI GTX 1050 Ti improves by 22% climbing to 70 FPS average remember we're roughly plus or minus 6% to multi player performance depending on settings Maps things like that so we're in good company at this point this test was it more for these lower end cards let's move it along to the higher end with 1440p resolution and the highest graphics settings and FXAA the 1080i is knocked down from its frame rate cap bumping numbers to 105 FPS average losing about 38% from the 1080p resolution things are still plenty playable though scaling reasonably through the GTX 1080 at 82 fps and Vega 64 at conveniently 64 FPS of iluminati isn't yet confirmed consider also that the GTX 1080i leads vega 64 by 64 percent jokes aside rx Bega 64 is roughly tied with the GTX 970 but the vega 64 card exhibits were same time consistency this is at least partly a driver side issue seeing as the performance here deviates from Vegas norm of being closer to 1080 performance and other games I'll check back once Andy is another update likely closer to launch here's a look at the frame time plot of Vegas 64 and the GT X 1080 FTW showing the 1080 FTW is more consistent frame time output remember with frame times it's not just that higher is worse but that more variance is worse you want the flattest line possible ideally closest to the interval of your display alternatively here's a Vega 56 frame time plot compared to the gtx 1070 is frame times showing a similar outcome back to the 1440p highest chart for a moment though we see that performance starts falling harder at the gtx 1066 gigabyte card gtx 970 and RX 588 gigabyte card players are more sensitive to frame rate may want to drop settings or resolution at this point here's a look at the drops settings option just to see how many more frames we can get out of going too high there's more room below this option of course but high and highest look close enough that it's not a huge loss to quote downgrade some cards are dropped from this chart as they were doing just fine at highest the gtx 1080 FTW runs an average FPS of 126 now and could hit 1440p with 144 hurts if dropping settings further the 1080 TI does this more readily Vegas 64 operates an average FPS of 106 with frame times reasonably consistent in this particular scenario it seems that Vegas 64 is more sensitive to one of the graphics settings in the highest preset which is something we'll explore later on under this video but also more as the game iterates with high frame times improved to a point of being consistent enough that most users won't complain though average FPS does need to climb closer to the 1080 to remain at least halfway competitive either way here's a frame time plot of the 1080 and Vega 64 showing improvement on Vega 64 bridges with highest settings we're stuck in a wait-and-see cycle on Vega right now with the present setting being wait to see if the new drivers chip and improve things as for the other cards back to the 1440p high charge performance remains above 60 fps across the board with low-end frame time performance also reasonably boosted all these devices could handle this configuration for the most part perhaps with the RX 570 needing a slight settings reduction to ensure a high baseline and/or variable multiplayer play one item of note the RX 588 gigabyte card improved from 38 FPS average at highest to 75 at high which is nearly exactly a two times performance leap the GTX 1066 gigabyte card saw a leap of 62% from highest to high which although not as impressive it's still a massive jump when considering the visual quality difference ie minimal if using AMD Hardware it's becoming clear that running at high settings is preferable to highest after we get through the rest of these charts we'll look at the impact of SSAO on performance to determine if that's the culprit for the big hit these cards are taking here's 4k with the highest graphics settings and FXAA we only have a few cards here go figure it's intensive and see that the GTX 1080 is C 2 is playing decently but not as well as we'd like our multiplayer match has had us at about 55 FPS average in this configuration but dropping a few settings the high would help improve to a more playable 4k level the GTX 1018 on TI and Vega 64 are both having trouble with these settings here's the last of the comparative charts high is more playable at 4k especially with a 1080i FC to CNN immense performance uplift of 71 percent from highest and the Vegas 64 card again nearly doubling and its performance the highest preset is clearly taxing in ways which may not be appreciated particularly in multiplayer matches when compared to the frame rate reduction in exchange this is true for both Nvidia and ante hardware but is more noticeable at 4k probably because of the higher pixel count there's clearly a large performance leap in some of the spot checks when jumping between high and highest settings with Nvidia Hardware we're seeing 60 to 70 percent for the GTX 10 16 some of these tests with AMD Hardware we're seeing 90 to 100 percent 2x increase and saw the benchmarks by going down too high from highest this is something we'll explore more in a moment but Andy's frame time performance particularly and markedly improves when going to the high setting where in the highest it doesn't really do too well to compete with Nvidia and high it's really not bad so just if you're an AMD owner take note you'll probably want at least in the beta for however long as last left of it at this point you'll want to use high rather than highest if you're having issues with frame times if you're seeing stuttering for instance or if you're just not happy with the frame rate things may change as the game and was mature but this is what we know for now in the interim there's clearly a benefit moving between settings so let's take a look at some of those specifically with the rx 580 and figured out which set of graphics options has the biggest impact this chart shows the RX 588 gigabyte card at 1440p using highest fxaa as baseline and the high preset as our target number which is significantly higher in performance we toggle the few settings between highest and high ultimately finding that depth of field settings are responsible for about a 38 percent performance hit on their own as switching to high boost FPS from 38 to 62 average between the highest and the high depth of field configurations 3d ambient occlusion is also responsible for about 13% of the performance hit on its own moving us to 44 FPS from 38 FPS average if you'd like to play on highest settings without sacrificing too much performance these are the two options immediately worth tweaking they seem to presently impact AMD the most noticeably including the vega card and the RX 500 series that we tested here and all that should give a pretty good idea of a few key things first of all because this is a beta because drivers may change for both Andy and Nvidia destiny 2 may further develop the thing to take away here other than performance today if you want to play today is the variance between multiplayer and singleplayer we'll look again when the game launches but for now it looks pretty consistent and that's a good thing for benchmarking purposes and if you want to compare a single-player forints and then the difference between graphics settings particularly high and highest with the biggest gains coming out of dropping to high there's a pretty big difference there and quick side note when talking about a single player we're really just talking about the campaign it drops you into when you start the game technically that's still multiplayer but it's a campaign not competitive multiplayer that's what we're going for here so anyway competitive multiplayer and the campaign are pretty close performance the cards in general at 1080p with high settings everyone we tested was fairly playable at all over 60fps I think maybe exception of one and going up to 1080p highest cutoff a few of the lower end cars but 10 65 80 and up can kind of do it you might want to drop some settings on the lower ones for the rest as long as you're at high settings you're probably okay for most of the modern cards we didn't get to test the gtx 960 AMD cards for this one but we'll consider them once the game properly launches and more drivers are out but for now that's your introductory look as always if you want to help out with this type of testing you go to patreon.com/scishow to donate directly and join our discord for patreon members and then gamers Nexus squarespace.com to pick up a shirt like this one this is the GN Graff logo shirt subscribe for more especially the CPU benchmark coming soon thank you for watching I'll see you all next time aside from that one zone where was really geometrically complex but