Gadgetory

our first Final Fantasy GPU benchmark is pretty unique for a few reasons one of them is that it has 1440p benchmarks the utility does not officially include the 1440p resolution but you can make it work through command-line we use scripted automation for all this testing so it should be pretty damn consistent and accurate run to run it's all done by automation so no real room for error there we studied the game completely for the benchmark portion to understand what we were testing I have a separate video on that we have a graphics analysis in this video that talks about specific graphics techniques that the technical artists used to help all of us understand why this game is as taxing of GPU hardware as it is and then final two unique points we have a Titan V in here just because why not it's been a month since we saw it and also I'm gonna occasionally refer to it as Final Fantasy XV just to annoy you all before that this video is brought to you by thermal Grizzly makers of the conductor not liquid metal that we recently used to drop 20 degrees off of our temperatures thermal grizzly also makes traditional thermal compounds we use on top of the IHS like cryo not and hydro not pastes learn more at the link below so let's start this out by talking about what they're doing with the benchmark benchmark is only three point seven gigabytes it doesn't contain the whole game it's not fully representative of the entire game however it's six minutes long and it takes huge portions of the game in the open world environment so we actually get a really good look condensed at the entire game as a whole despite it not being necessarily fully representative everything in the game this means we can analyze the graphics quality because it's actually really damn demanding and the visuals of the game are pretty impressive so you look at things like the metallic paint where you can see paint specs and one of the car paint jobs it's actually pretty cool stuff and we'll talk about more of that in a minute before getting to that there's some information as well on the benchmark specifically how Square Square Enix is uploading the data from I don't know if it's aggregate data from users or how they're collecting it but basically there is some form of aggregate pooled data on their website and depending on how you test your device there could potentially be some poisoning of data there because it's not necessarily fully controlled those tests could be run on various configurations with various CPUs for example which would skew the results so we are running the test in a controlled fashion and we also have 1440p benchmarks in here just to give a bit of an extra but let's move into some of the first major items that we notice with performs a quick note confirmation was received from Square Enix on the settings we talked about this in a previous content beast where we were uncertain if the graphics settings were the same between high medium and low across all resolutions Square Enix did confirm for us that the graphics quality settings remain the same across all resolutions so the very first thing to point out is that you're gonna see interesting significant differences between AMD and NVIDIA and the GPU charts and this is particularly at high and medium settings you'll also see differences between same brand devices across these two settings so this is partly because of the frame time variance that we can pull up a chart now that shows the frame times of an Nvidia card at high and medium using the settings shown in the chart we're using an Nvidia card here because this is a game works game and Nvidia's had the most access to it presumably so to show that they have frame time variants reveals that the game doesn't run perfectly even on nvidia hardware who have optimized for it the most because it is using game works options so frame time variants here is far greater with high settings and frame to frame intervals become looser and more stuttered this causes visible hitching so to speak during gameplay and exists even on 1080 T eyes it's particularly bad during the first run of a benchmark but is ever-present with high settings moving to medium largely removes the issue as shown in the frame time plot despite being an Nvidia supported title both Nvidia and AMD suffer in the frame time Department specifically with high settings so as we go through the benchmark charts keep a close eye on Vega 56 and Vega 64 positioning relative to each other at high and medium let's get into some of the graphics analysis of the six minutes of the benchmark shows us and why those six minutes are so taxing one of the most immediate and important items to point out is the tessellation in the game so between Andy and Nvidia they both handle geometry differently Andy has expanded its geometry pipe in recent years but NVIDIA maintains an advantage in dealing with heavy tessellation something that's been true since the crisis days at fifteen seconds into the benchmark we noticed that Square Enix used a heavy amount of tessellation in the ground and terrain rather than using a traditional normal map for the ground this adds perceived depth to the surface by allowing displacement to increase apparent detail via tessellation and tessellation is basically a doubling of the triangles on the graphics hardware so four becomes eight and so on this makes the service seem higher resolution and it's also why we get the added bumps a few of which we can highlight in the video production the card doesn't conform to the surface from what we can tell but it's more geometrically complex of a surface and it makes it more interesting to look at the actual value however is debatable geometric complexity like this cost a lot and if your device and the or Nvidia is struggling with heavy tessellation we'd assume you'd be able to separately disable this in the final game there are also options in the driver settings that where you can tune this down interestingly at three minutes 58 in the video it appears that the ground is no longer tessellated we think that Square Enix has switched to a traditional normal map here we're not sure if this is because it's a known combat scene with a lot of other complex elements to render or if it's for some other reason let's rewind to some other impactful settings at 27 seconds into the benchmark we noticed some high quality reflection techniques being used by Square Enix on the car the door and the mirrors of the truck are dynamically reflecting the environment we don't think these are using screen space reflections because you can see clouds and other background elements dynamically reflected in those services at 50 seconds we notice that the trees are using flat alpha textures instead of particles or some sort of game works plugin this is a bit of a primitive way to do things but as cheap in terms of resource cost five seconds later we pass a truck with extremely detailed textures as seen on the front of the vehicle and this is a major contributor to vram consumption that penalize is for gigabyte GPUs as you'll see in the charts and a bit there are high quality textures in this game there's also some motion blur going on and we further noticed that the truck leaves behind some ghosted lines that slowly fade something we saw again at three minutes four seconds where the fishing pole leaves similar ghosted lines at 56 seconds just after the truck you can see some highly noticeable limited shadow draw distance poppin in the background where the environment elements and the shadows are getting drawn as we approach them at 1 minute 54 the chocobos are using traditional alpha textures for fur as opposed to hair works or other game works effects which we know are bundled in because they used on character model hair for the main characters the tail feathers do use some sort of advanced physics calculations when the chocobo runs or moves but nothing special for the fur as far as we can see at 2 minutes you can see soft shadows cast by trees and it's noticeable by the way the shadows soften as the leaves get further from the ground as opposed to the base of the tree that's closer to the ground and uses a harder shadow this might be part of Nvidia's frustum traced shadows but it could also just be an older soft shadow technique at 2 minutes 33 seconds there's a great example of screen space reflections if you don't know what that is you will now we can demonstrate this by looking at the tree leaf that includes the background mountains the game is using that screen space to calculate reflections of the mountains onto the lake but the occlusion results in missing information in screen space so the reflection is inaccurate and incomplete and you get that weird reflection effect on the lake the alternative to this would be planar reflections but the hardware cost is too great to use in most real-time scenarios 3 minutes 53 seconds is also an interesting scene as Noctis warps across the screen we can see the background bridge refracting and noctis's mesh similar to what you'd see with a straw in a glass of water for instance this is another advanced rendering technique that has some physical hardware cost to it will cut that there but keep an eye out for our full graphics analysis coming soon subscribe for that we're trying to do more of them for now though we've established why the game is as graphically intensive as it is and why it's so taxing on the hardware let's get into the charts as always you can check the article link in the description below for more information on this and for the bench platform our 1080p high chart will start us off as that's likely what most of you are interested in the Titan V has some immediate difficulty with the lows here more than the other devices but it's also not really something we're expecting anyone else to use to play games we just threw it in because we had it moving on to actual gaming cards everything up from the Vega cards and up do reasonably well and averages but the frame time hit at high settings introduces occasional stutters or lawn intervals between frames we showed those frame time plots earlier this is the real impact of that this impacts AMD and NVIDIA alike though AMD does take a bigger hit as illustrated on Vega versus these 0.1% lows of the neighboring 1070 and 1062 we're talking 9 FPS versus 15 FPS 4.1 percent lows so it's not really a major victory for NVIDIA nothing to brag about at this point what we really need is the ability to custom tune settings so that we can find out which one is tanking the frame time performance and causing those long frame to frame intervals until the game launches though we won't have that so we want to 1080p medium next this settings configuration would be ideal for something like a gtx 1066 gigabyte card with the RX 500 series high-end not doing too poorly here either you can use either and probably be happy with it especially once you can actually tune the settings manually vegam 56 does reasonably well despite its frame time consistency and manages a 77 FPS average the card is still outperformed by the 10 70 by about nine point seven percent but we'll see if a.m. these future updates do anything to help with this before getting into 1440p we also ran a few low-end devices at 1080p low settings results are on the screen now the 1060 predictably leads managing 80 FPS at low and this device can handle medium settings fine so no surprise here the RX 574 gigabyte card is struggling in the frame time Department once again but manages a FPS average respectively the 1050 Ti places similarly to the AR five seventy both at about forty nine FPS average the RX at five sixty 16 a Cu card the original one that was shipped not the later ones with 14c use is about 10 FPS behind the 1050 TI OC it's not actually overclocked MSI just calls it that it's one of the cheapest 1050 T is 1440p benchmarks are something we're proud to someone uniquely offer because we scripted our tests and automated them so we're able to force the games we run at resolutions not included in the built in benchmark that includes 1440p at 1440p high settings obviously without the ability to customize the options the gtx 1080i exceeds 60 FPS average and manages to hit 72 FPS throughout the lows at about 49 and 22 fps looked better than previously but 22 is still dismal when compared to 72 as your average this is a problem with the high settings primarily the 0.1% low values here are still suffering this is repeatable on all devices when tested with high settings for the graphics although the average framerate may still be 72 FPS you'll see visible stutters and hitches on occasion particularly when loading a new area for the first time at that boot or in the first time in a while we'll need to wait for the final game to see if this gets ironed out the 1080i leads the 1080 by 28% with a gtx 1080i operating at 56 FPS average with similarly painful frame times the GTX 1080 leads the next card the 1070 TI by 11.8% and there's enough of a gap here that overclocking could more or less close it even when I also overclocking the 1080 the GTX 1070 manages a 46 FPS average with the Vega 64 Strix card operating a 39 FPS average pay close attention here to Vega 56 64 and the gtx 1070 right now the 1070 is leading even vega 64 which probably shouldn't be the case but as we move on to 1440p medium you'll see what's going on so here's where it gets interesting then at 1440p high to recap the 10 does have any leads at vega 64 by a sizable 16.5% and it leads its direct Vega 56 competitor by 30% that's absolutely massive considering the are much closer together in other benchmark titles and at Vega 56 often even outpaces the 1070 in like-for-like testing at 1440p medium things change the gtx 1070 now runs at 60fps average with the vega 64 cards surpassing it at 63 FPS average they give 56 still trails at 55 but previously we were about 30% ahead with the 1070 over vagin 56 and with medium settings from high the 1070 is now just 7.3 percent ahead of Vega 56 that's a big reduction as predicted in our pretest video the gap was closed almost entirely by dropping a couple of settings potentially gameworks options for which a and he hasn't yet optimized because medium disables all game works options that would explain a lot here it could also just be general tessellation and LOD or view distance settings which would feed a lot of geometry and primitives to a card that has historically compared to its competition struggled a little bit more with both still the 1070 TI leads both of Vega devices and both Vega devices still struggle on the frame time front here's a plot of the 1070 T is at frame x for the medium runs mapped up against Vega 64 s frame x additional spike eNOS and the AMD card presents itself in this testing we'll have to wait and see if Andy's Final Fantasy XV driver is released later closer to the game's launch will fix any of this and he has told us they're not planning to release drivers for this benchmark today but they'll address it once there's a real game to address at 4k high settings the gtx 1080i operates an average FPS of 46 with lows at 27 and 1212 pretty pretty bad again this frame time deviation is derived specifically from the high settings until a point at which we can manually toggle those settings we can't say for sure what specific options are causing that behavior none of these devices are particularly well equipped for Final Fantasy 15 at 4k high in its present state but full graphics customization options should help the Titan VIII for what it's worth operates the highest average by about 21% but it also has worse frame time consistency a lot worse and it's to a point of being actually noticeable in gameplay I'd rather have one of the lower end devices with higher 0.1% lows to play this game than the Titan v4 4k medium settings the GT X 1080 Ti struggled to maintain its 52 FPS average with 1% lows dipping down to 41 FPS frame times aren't as much of a concern here as they are with high settings as illustrated by the close proximity between the 1% and point 1% lows and all tested devices except for the r9 fury X which has two times worse 99.9% I'll frame time values the GT x 1080 predictably lands second but it's a shallow victory 40 FPS average marking the 1080i about 34% ahead of the 1080 the RX Vegas 64 Strix card puts in a good run with its 36 FPS average comparatively led by the 1080 by about 8% dollar-for-dollar assuming GPU prices were a thin anyone abided by that would be a direct comparison low-end frame time performance is a bit worse on the Vegas 64 card but this game isn't particularly 4k friendly anyway the fury X isn't too far back from Vegas 64 and average FPS but drops the ball with 0.1% little values this is probably at least in part due to the frame buffer size because you can look at the game see the texture sizes and generally assume that it's using a lot of memory so over the conclusion there's a lot more that could be said here we'll try and outline some of it in the article below that said those testings been going on for 13 hours at this point I think so actually more than that so it's been a long day of testing we have lots of content on the website and on the channel already if I've left something out sorry we'll get to it eventually but it's I mean it's it's 4:30 a.m. cut me cut me some slack so we've been testing since the benchmark came out we have several articles on the site we also have CPUs to go still and we started that testing just need to finalize it if you feel like there's something really specific that you'd like us to look at for CPUs you still have some time to get that request in because we are still working on getting the benchmarks going got the methodology figured out need to get the tests running so that's it for now subscribe for more make sure you catch our graphics analysis and CPU coverage and as always good at patreon.com slash gamers Nexus tells out directly where I've been hanging out in the discord talking with everyone as we are running these benchmarks live and finally go to store that gamers access net to pick up a shirt like this one or one of our other products thank you for watching I'll see you all next time