Gadgetory

we already gave Metro Exodus the full run-through for graphics analysis and looking at our TX on versus off high versus ultra you can check our other video on that but what we haven't published yet other than today is the benchmark performance so today we're looking at a lot of performance characteristics of Metro Exodus this is wide-reaching we look at dx11 versus dx12 scaling how do they perform versus one another look at dx12 scaling across multiple settings like with the 2080 CI we got built-in benchmark versus in game and manual benchmarking and which one is sort of more reliable and then we're also talking about our TX scaling across all current r-tx compatible 20 series devices so that's our roundup for today's benchmarks with Metro Exodus and let's get started with the built-in versus the in-game benchmark before that this video is brought to you by the EVGA CLC 280 liquid cooler people ask me how I keep cool during the summer with all this hair well I've tried a lot of different products and if you do exactly what I need many of them cause tangles or worse EVGA CLC 280 helps keep my core temperatures low they're in hot benchmarking sessions the CLC 280 is price competitive and focuses on performance for value offering a 280 liquid cooler at an affordable price get yours at the link in the description below hair mounting kit sold separately getting into this one so we noticed a couple of really important things here that should be pointed out first you have to restart between all changes in Metro Exodus it does have some issues with applying the shape this is really not abnormal for game testing but Metro Exodus in particular it has some some odd performance behaviors if you don't restart between even resolution changes sometimes so this isn't a hundred percent of the time there but the randomness of the performance anomalies and by anomalies I mean sometimes a little hard lock to 60fps and it's just it's not because of B sync you can force vsync off and global settings and env CP whatever doesn't matter sometimes they'll just hard lock and there's no fix for it and that can be from not restarting between changes they can be from other things too that are less obvious like having desktop resolution or monitor resolution different from game resolution so if you have an FPS lock that might be it it can be from scaling and windows stuff like that it's got some weird bugs and quirks in there and then separately we did have issues with the benchmark occasionally locking to an odd 62.5 fps very specific number that was resolved by walking away and coming back later so we're not sure why that CUF fixed but it did so this this already was off to a start where there's that kind of benchmark anxiety of can I even trust my numbers because I don't know if I can trust the game we got it to a point where we trust it unfortunately I couldn't tell you how we got it there just it took a lot of just weird troubleshooting effort and one of the things worked so anyway that's what we're looking at also the first benchmark run is often inaccurate it scores lower than the rash we'll talk about that today too so let's let's start off here with that built in versus in-game testing and show the differences between the two one of the most important things to look at with major game launches is whether the built-in bench mark is actually reliable with final fantasy we discovered that the game had improperly called objects to the extent that they affected performance from miles away horribly inaccurate to actual gameplay with other games like 6 we found that the turn time benchmarks are actually pretty realistic the gameplay for metro Exodus will start by checking actual in-game performance versus the benchmark scene we're starting with only the 20 atti at 4k resolution scaling across low to extreme settings and with varying r-tx configurations we found that the built in benchmark operated at only 84 FPS average with the 28 ETI at 4k and lowest settings with r-tx off mind you whereas our in-game testing posed in 152 FPS average performance results pretty big difference there as always this is tested multiple times an average for parity and these numbers are very consistent at this point at medium settings Authority acts off we found reports to decay to 72 FPS average for the built-in test and 118 FPS average for the actual gameplay this further drops to 60 FPS average for high settings or the built-in benchmark and 94 FPS average for the in-game test for the next data point we have to highlight something important with ultra settings and r-tx off we're still at 87 FPS average which is technically better than the built-in benchmarks 84 FPS average at low settings so you can kind of visual line there from one to the next and see how big of a hit there is for running the benchmark versus actual gameplay if ever there were any question as to whether the built-in bench work is an accurate depiction of gameplay this answers it by three entire drops and graphics quality settings for extreme settings we hit 70 FPS average without any arch acts enabled and 40 FPS with all graphics settings completely maxed you'll see that at the end of the charge but as we descend in between those two we eventually see that the r-tx 20 atti hits 53 FPS average with r-tx on high and graphic set to ultra with in-game performance vs. 36 for the built-in tests we then hit 48 FPS averages our TX ultra adop about 9% from high and we show the differences between high and ultra visually and qualitatively in our previous content piece but we'll show some of those as refreshers here and on average the differences are minimal between high and ultra but you do have about a 9% hit so all of that sets the scale for us we'll show some more footage from our qualitative comparison while explaining some of this as for what it all means it's pretty straightforward as long as the card to card scaling remains equally spaced in the built-in test as it is in the game for example at 20 atti retains X percent lead over a Radeon 7 in both tests where X does not change then it is acceptable as a synthetic gauge of performance it is not however useful as an absolute gauge of performance for this game so anyone wanting to play Metro Exodus specifically should be looking at in-game benchmarks not the built in benchmark it is you'll end up buying something way higher-end than you actually need and you might not even think it's possible to get what you need if you look at the in-game benchmarks instead for comparing two cards relative to each other though the built-in test is still useful it just comes down to having equidistant relative scaling between devices as long as that exists you can use it as a synthetic comparison it's functionally x bi or firestrike at this point so useful in so far as determining the distance from one car to the next as a percentage but not as an absolute FPS value so we're gonna stick with the in-game testing for the rest of our work because it's more universally accurate and let's look at some more detailed performance numbers next except for this one all tests in our bed today are with directx12 enabled as its required for DXR we still wanted to test the x11 though as sometimes game developers will lazily rap the more abstracted api or fail to really leverage the x12 worst cases we see some performance loss for this one we took the RT x 28 e non TI XC ultra and tested it at 1440p in ultra settings with RT x again disabled we still aren't heavily loading the cpu here so this is very much a GPU bound benchmark with dx11 we measured performance at 114 FPS average 91 FPS 1 % low and 89 fps 0.1% lows the DirectX 12 in-game test put us at 112 FPS average 84 1% lows and 79 fps 0.1% lows that's more or less tied but we need to look at frame times to better determine if we're operating outside of variants as always frame time plots show us frame to frame present intervals to get the best picture of per second experience of gameplay lower is better in frame time but consistency is more important than just being lower for this one we see that DirectX 12 testing plots us at starting around 8 milliseconds then climbs to around 10 milliseconds eventually hits 12 millisecond frame times the more important aspect is that the spike eNOS of the frame time is measured swings up more than DirectX 11 does as you can see here these are both completely playable in the frame time variants never exceeds 8 milliseconds with dx12 enabled which means that the average user really won't know the difference the x11 to be fair does have tighter frame time consistency and is objectively better it's just that when it comes down to it the objective superiority doesn't really matter in this context it's measurable but it's not perceivable to the end user for a seems to have roughly equal performance in the x12 and dx11 maybe slightly favoring the x11 on average now that we've whittled down testing to just focusing on DirectX 12 we can look at graphics settings and r-tx scaling more granularly for in-game testing performance scaling from the 20 atti @ 4k shows reasonably strong results at 4k ultra with r-tx off giving us an 87 FPS average this grants high settings with our TX off a gain of about 7 to 8% against ultra settings or ultra a gain of about 25% against extreme settings we think that ultra is the best middle ground on average at least for hire ish and devices considering that extreme shows severe performance fall-off medium is another major step where there's a severe change for medium to high but texture and shadow quality really start to become noticeably bad with medium settings for ultra settings with r-tx off our 87 FPS average is 65 percent higher than older settings with r-tx on its lowest setting which anybody calls high presumably because nvidia does not want to dirty its brand with an option named low even though it's basically low and high in this case if you prefer to look at it the other way it's about a 40% performance drop from r-tx off and going to ultra drags us down an additional 8% from our TX high rendering even in game frame rates undesirable the impact of our TX appears to minimally be about 40% FPS a loss - 45 % loss at Ultra as for what that gets you again we'll show some of our footage for our previous technical analysis of the graphics quality where I was joined by Andrew on our team to talk about the qualitative look at the games visuals the end result is that some scenes particularly those outside where light bounces off snow for a less faked global illumination we get additional contrast some more gradual shadow gradients like across the chest of one of the characters on the front of the Train for instance and in other scenes like the actual namesake Metro of the game Metro Exodus there is functionally zero impact from RT acts and visuals to the point that it's just confuse you won't wonder if it's broken because there's still 40 to 50 percent hit the framerate in exchange for nothing you can learn more about the visual appearance differences in our previous video on the channel and of course subscribe if we do any follow-ups but let's move on to another test a quick note before progressing further we noticed that the first round of a benchmark would produce numbers sometimes 8 to 10 FPS lower than subsequent benchmark runs and this is for in-game testing so these mostly averaged out when considering multi pass test approaches but just be aware that your first exposure to a new area may be at a lower frame rate than repeated or prolonged exposures to those areas the numbers that have been on the screen now sort of demonstrate that we can see a bit of performance fall-off on that first run and then some crack shortly after that this doesn't always happen but it happens probably about 90 plus percent of the time our t acts are only testing is next for this because we're specifically looking at our TX scaling card to card the charts will be limited to only compatible 20 series devices at 1080p ultra with our TX set too high because Ultra just seems kind of pointless the 2080 ti tops the chart at 138 FPS average with lows at 99 FPS and 94 fps 1% in 0.1% these are good numbers overall and reasonably consistent so frame time pacing isn't suffering here that said compared to our TX off numbers we are still obviously taking a large performance hit the 2018 non TI ends up at 114 FPS average allowing the 2080 ti elite of 21% and leading the are TX 2070 s 90 FPS average white 26.6% the 2070 then leads the 26 these 80 FPS average by 13% this gap is much less impressive than the previous two and considering our review of the RT X 2060 we know that an overclocked 2060 could somewhat easily catch the stock 27 a performance granted you could also overclock the 2070 but it's not hard to get baseline stock performance out of an overclocked 2060 1440p is next for these settings the RT X 28 e TI fe ends up at 101 FPS average posting a loss of about 13 percent from the previous chart lows are still tightly timed and within reason the RT x 2080 runs at 77 fps average granting the 28 e TI elite of 32% this impressive gap should widen as resolution continues to increase the 27 t ends up at 60 FPS average and the 2060 isn't far behind demonstrating significant value when considered against the 2070 again a quick overclock would balance things here but the distance between a 2060 OC and a 2070 is much less than a 2070 OC and the 2080 keep in mind further that drop-in settings too high from ultra for the preset not for the RT X preset would give us some performance back although RT X is responsible for eradicating the majority of framerate the last chart is almost embarrassingly barren but to be fair it's really not our fault at some level notably below 30 fps it sort of stops becoming to test the lower end video cards at 4k ultra and with our TX set to high the r-tx 20 80 TI leads the 20 80 by about 36% this again expands its lead as the r-tx 2080 begins to fall behind from the sheer pixel throughput but both devices are struggling in the face of our TX enablement still it's far more playable than the pre-built benchmark scenes dropping settings would make this a bit easier but it makes the most sense to just disable our TX altogether if in desperate need of a higher FPS at 4k for example a 20 80 is perfectly capable of 4k with a reasonable FPS in this game you would just have to turn our TX off for this game the graphics difference qualitatively there is a pretty significant difference and I speak about this with Andrew in our other video where we look at the graphics quality visual quality is there a noticeable difference of perceivable difference in the game when you toggle our TX on versus off so for high vs. ultra versus off there's really not that much of a difference there is about a 10 percent performance hit by going from high to ultra you can see the difference if you really look at it but at some point if you have to be sitting there with your face three inches away from the monitor looking at each quadrant of the screen to find the difference the difference is irrelevant so for that reason we would generally suggest if you want to use our TX going with high instead of ultra and just taking the extra frames because that's going to be worth more to you than the often in perceivable differences of ultra now again they do exist but it's at a level where if you're actually genuinely playing the game which we weren't we were looking for these differences you're probably not on average going to notice so that's the first item of note the second one is that in some scenarios in the game like again the metro we talked about the actual metro it's uncommon to see these differences materialized but the performance difference does materialize so where you won't notice any visual impact you will notice a framerate impact to the tune of about a 65 percent improvement with our TX disabled or if you prefer to calculate in the other direction because stat math is funny and sometimes misleading it's about a 40% reduction from the baseline off so you weight how you want depend on how you want make the numbers look I guess but that's what you're working with but that's not the only type of experience in the game the Metro of course there were significant visual differences in sort of the overworld like the train hub level where you've got that ray bounces off of the snow they hit characters and if you don't know what r-tx global illumination is well that's what the toggle does when you enable RTX in this particular game it's the first implementation of our TX GI global illumination is most commonly showcased as it was with Nvidia's presentation by using sort of multicolored objects within a room maybe you have a big area light above it and you can get some sort of like color casting where when the light shines down on the objects or on a wall or whatever you'll get almost a radiative glow off of the object of the wall and we'll show some footage of one of those just to illustrate what it does and and what you're looking at so that would be a global illumination demonstration that's what you get with Metro not does some other things too like again bouncing Ray's off the snow to illuminate the underside of things or to illuminate characters where normally you would have to sort of fake the graphics by putting in maybe a hidden or an invisible light source on one side of the train for instance to create the illusion of global illumination without the cost of really doing it in a real way so r-tx is an interesting way for developers to achieve global illumination or other effects without putting in manual effort is it worth it well most the audience is still playing without r-tx even if you own the cards you might drop it for performance so it'll be a while before we see this really wide spread and a lot of developers will probably on average still be using those the cheats and that's really what game graphics is it's all about cheating the graphics to make it look as real as possible because if you went for true realism it's just not going to be something you can render in real time so r-tx finds a middle ground there and sometimes it works incredibly well like in some of the outside scenes sometimes it's completely nebulous and up for debate like inside the Train or actually the front of the Train personally speaking subjectively I prefer to RTX off I thought it looks better now that might be because I'm trained to looking at games the way they're made today on average but I thought it looks better so I would turn it off and then there are other place where it's just it's a very clear there's no benefit like in the Metro so there are three very specific scenarios of objectively just superior and then completely subjective you might not like it and then there's no value at all and because of that mix of things plus the performance we're seeing we wouldn't necessarily let me rephrase that we would not recommend buying an RTI X card just because you want to use RT X yet now if this starts to become more widespread if it doesn't become another abandoned and video technology it may well be worth it if you're a developer if you're someone in graphics if you're in 3d art there's a whole host of reasons you could argue RT X is worth it but if you're a gamer then metro on its own is not really a good enough reason but either way you have some benchmarks now for Metro so you can see what it looks like RT X is at least starting to get implementations in games we can't fault in video for that because we did fault and video for not implementing in games previously so they're moving in the right direction RT x does provide some value sometimes it's not enough quite yet but it's clearly finally getting some spread so we'll see how it develops there's a lot more to look at here we'll be back not too long from now with more RT x content to kind of look at how it's advancing and if this is something that we need to pay more attention to for the future so that's it for this one thank you for watching as always you can subscribe for more we'll probably have one more content piece on Metro here it depends on how how drivers and things like that work out but we might return with one more piece we have an idea on so subscribe for that one you go to store documents access net to pick up for example our brand new medium mod matte which is at time of filming in stock though they're selling quickly so you might be on backorder buy but we'll get more in stock shortly and that's it for this one I'll see you all next time