Gadgetory

apex legends may have launched embroiled in controversy about not being Titan vol 3 but that's been apparently resolved because it's now one of the most played games on the market and is often top two if not the most played or most viewed anyway on twitch TV so today we're looking at apex legends for GPU performance we're testing a wide range of video cards and we'll be looking at performance study in different areas of the games a testing practice mode versus multiplayer mode different areas of multiplayer mode and then ultimately determining what's the best test path for GPU benchmarking this gets fairly involved because apex Legends is a higher fidelity first-person shooter game with a high player count with a really big map long view distances and geometric complexity that can impact cards heavily so let's get into the testing today to look at video card performance and how they all stack up against each other before that this video is brought to you by the coarser 1i 140 compact gaming PC the corsair 1i 140 is a small form-factor PC outfitted with a 9700 K RT x 28 e 32 gigabytes of RAM and a 480 gigabyte nvme SSD all housed within a 2 millimeter thick aluminum chassis the corsair 1i 140 is a 12 liter system fit for desktop use with the same sized I 160 counterpart with higher-end parts learn more at the link in the description below apex legends is a battle royale game if you hadn't heard it's also sort of titanfall in that it's using the same universe and a lot of the same assets or similar ones the same engine even but it's not literally titanfall there aren't Titans falling from the sky instead its players it's more of a pilot fall it's also not titanfall proper it's not the same sort of wall running mechanics and things like that so if you're familiar with titanfall this is different this is one of the more annoying titles test because origin for some reason still has a 5 hardware change limit to testing the game so if you change your video card five times in the amount of time allotted it will restrict you from playing the game any further even though it's a free-to-play game so we end up just making a lot of accounts sorry EA but let's get into the performance testing and and we'll start off with sort of the study this is what we do when we try to figure out how we should even test the game before committing to a full benchmark pass the most important thing to do before embarking on testing dozens of configurations is to establish some level of reproducible results within the game multiplayer is clearly where everyone is playing but the map is huge it has desert regions it has grassy and mountainous regions it has a river with semi flooded buildings that are very graphically intensive and there's the initial drop in from the sky with a fuller view of everything framerate is all over the place in this game if you're properly playing it which is what makes it functionally impossible to benchmark with full accuracy to absolute numbers just watching through clips in our video that you're probably seeing now you can see how varied the environments are and the best we can do is establish a reliable benchmark that gives accuracy relative to performance scaling between cards so let's start with a study of 4k performance to see consistency from match to match well look at frame time plots after we get through the frame rates graphs that we're looking at next the 1080 TI and 4k settings at high had us observing an average FPS of 47 when dropping into the match with a framerate primarily driven downward when looking over the landscape at a glancing angle but wider views of the landscape and horizon drove FPS down to about 47 FPS average for five minutes in the hills and the burned forests we monitored framerate at about 60 FPS average with lows at 45 fps 1% and 39 fps 0.1% we have some footage of that as well but this area isn't really that demanding it's much less demanding than the river area that is later is elsewhere in the game so this was just coincidentally 60 FPS average vsync was off adaptive sync was disabled finally testing with practice mode instead of multiplayer we see that our framerate shoots high we're at 73 FPS average instead of 47 to 60 FPS for the range sort of within reason but it's clearly not realistic this is a 20% performance gap and even ignoring that there's still a question of whether other cards would scale accurately in practice mode the answer has you'll see later when we prove it with some numbers is no they do not scale linearly between practice and multiplayer and so we ultimately abandoned practice mode and instead went for multiplayer testing with multiplayer as long as testing is conducted in exactly the same spot every time this fashion it's accurate from run to run in games game we tested this to the testing only deviates and becomes variable when testing in random locations and multiplayer at 1080p the initial drops plot between 126 fps and 137 FPS average with that difference primarily emerging from whether we look straight down or with a wider viewing angle of the horizon our two 5-minute test passes it covered the entire map plotting between 162 FPS average and 182 FPS average with a low is functionally equal between each in this specific game there's a lot of range and what that single what the map shows because it's one map for everything as opposed to something like black ops where you might have multiple maps that are more easily controlled practice mode also poses a problem though with even our heavier load practice testing landing at 206 FPS average this ranks it as 13% ahead of the fastest gameplay test not unreasonable really or 27% ahead of the slower test which is unreasonable for these reasons we're going with multiplayer testing in a controlled area with heavy load using the river region of the map this will give us data that best represents the real gameplay experience but remains controllable for variables and is reproducible in testing so this gives you both the absolute and relative scaling with only a few caveats in between that we'll discuss later finally getting into the results here's 1080p with multiplayer testing near the river part of the map where we see medium to heavy load without introducing one-off load inducers like grenades more on that after these charts further note that we disabled the framerate limit completely so we're not limited to the 144 FPS cap that is natively in the game the 28 ETI ends I've had about 161 FPS average with the Radeon seven card shockingly close at 153 FPS average we didn't trust these numbers looking at them they didn't really match any of our other scaling between the 2080 TI and Radeon 7 so we reran both devices again then reran with a different 2080 TI a third time just to be safe at the end the numbers repeated the Radeon seven card falls off hard as resolution increases something will show you momentarily and it's curiously counter to what one might expect with its memory bandwidth but also something that we showed as possible in our Radeon 7 review this game does seem to run well on Radeon 7 at 1080 P when compared to the 2080 ti we might you would think be bumping into some sort of frame limit on the CPU at 1080p but our peak FPS numbers during this test do post at 242 to 60 fps and above so those spikes won't be shaved too short if at all there may be some heavy draw call regions that are more CPU intensive but on average we are reasonably distant from the FPS bottleneck or at least concerns of it so we can just say that Radeon seven scenes to do well here the 20/80 trio ends up at 145 FPS average allowing the 28 ETI a lead of 11% with the Radeon 7 by a couple of percentage points the 1080 Ti is further behind the 20/80 than it has been in other titles ended up at 128 FPS average to 145 FPS average of the 2080 producing a gap of 13% favoring the 2080 and allowing the 1080i to match the 2070 in performance for this title it seems that an rx 580 with a light overclock ie and rx 590 it's the same thing or NRT X 2060 would be good modern-day options for purchasing for the mid-range - upper mid-range price categories the 980ti if you have one still holds on well ranking at 80 FBS average with 1080p high settings and the 970 seems to do alright remember that settings can be dropped below what we're running to further improve performance we're at high so you've got a couple of ticks there if you do need to up take it a bit the fury x ends up with frame time consistency issues pretty serious ones as does the rx 570 where both have low frame rates dropping below the mean distance against the average that we see in other tests 1440p has the 20 atti at about 124 FPS average so we see uplift of about 62% over the average FPS of our upcoming 4k numbers that we haven't shown yet the low performance is still consistent and good on the 28 ETI with frame times on average within a few milliseconds of each other the 28 ETI runs 12.7% ahead of the RT x 2080 which runs 3.4% ahead of the Radeon 7 card so we've now flipped in the performance rankings between these two and this is about within the performance range that we observed to the radio I'm 7 and our initial review as for the rest the 2060 seems to be a line of division against the lower end of the chart offering a good upgrade pathway for those wanting to play apex legends at 1440 key with high settings 82 FPS average with lows around 60 is widely acceptable performance for most people if you want more we do have other cards on the chart interestingly we see big performance hits on the fury acts again in the RX 570 where we think for the fury acts the limited vram might be causing some of that massive hit the frame time consistency dropping Louis 222 FPS your own percent in 36 F he has one percent so arms 5 to 70 experiences similar behavior here the GTX 970 curiously does not seem to encounter the same issue so less of an issue on this one we believe that it's so constrained elsewhere in the pipeline on the 970 that memory is perhaps not coming into play in quite the same way as it is for the more powerful fury X the card the 970 is having trouble enough running at even 30 FPS average 1440 view high so it may come down to architecture it may come down to driver optimization for the game but either way it's not playable with these settings so moving on 4k at 4k high the 2080 ti ends up at 76 FPS average ranked about 16% ahead of the RT x 2080 trio and 34% the head of the Radeon 7 the 2080 ends up at 16% ahead of the Radeon 7 card which is functionally tied with the 1080 Ti interestingly in this game we actually do see the 2080 placing reasonably above the 1080 TI in our launched a steam the 2080 was closer to parity with a 1080 CI in just about every game and sometimes falling a bit behind here the 2080 weeds the 1080 Ti and speaking of parity there the RT X 2070 isn't far behind the 1080i in this testing with a 1080 TI posting a lead of just 3.4 percent this matches what we saw in the practice tests as well that we haven't shown yet although we consider those tests largely invalid for wider performance analytics due to scaling limitations and limitations of issues shown like with the fury X versus multiplayer still interesting but not particularly useful if you want to get an absolute idea of how performance is the r-tx 2070 leads to 2060 by about 20% and at this point we'd probably recommend the 2060 dropped to much lower settings or to 1440p the 2070 does okay but in a fast-paced game like Apex Legends it probably also benefit from a slight settings reduction they give it these six struggles to keep pace here although it's frame time consistency and performance is good in this title overall so for that it gets praised the r9 fury x ends up behind vega 56 with low is dropping off to inconsistency potentially a results of those four gigabyte limitations especially at 4k the fury X also ends up roughly tied with the aging and once Kane 980ti which is just 8% ahead of the XFX rx 590 clearly these cards are not really meant for 4k gaming but some of them could be made to play apex legends at nearing 60fps with enough graphics quality reductions it'd probably be better to just play at 1440 though if you wanted to and if you were close enough like in the 50s you can drop down to medium settings instead of high and be within reasonable territory all of these numbers we just showed are pretty accurate for multiplayer use on average so you can use these as an absolute measure of performance this will of course become less true if patches launch of the game is optimised if performance is fixed on the developer side in any way or if drivers changings but for now they can be used for both absolute and relative performance scaling numbers and they can be used as relative performance scaling numbers as the game gets patches this becomes a bit less true in the absolute most load intensive scenarios we can find but for the vast majority of the multiplayer experience the numbers are good as a representation of performance one example of where our numbers would no longer be reliable as an absolute metric would be if involved in very heavy combat where multiple grenades are going off particularly the incendiary grenades as an example those incendiary grenades that we tested in practice mode where you have an infinite amount of them by the way would drop framerate by roughly 23 percent when detonating several in sequence like three of them in a row that's a big performance hit and it isn't accounted for in our charts so keep that in mind for decision making if you want to make sure that things hit a specific FPS target always in all scenarios including multiple incendiary grenades and close proximity factoring another 20% drop to FPS in those severe scenarios to be fair this isn't that common and it requires a lot of inventory of those grenades multiple people and a lot of combat but it can absolutely happen so it's an important factor to know about finally here's what a practice mode test would look like let's first firmly know that this is sort of invalid data will mark the chart clearly that way - since people will otherwise just jump to this point pasted everywhere without listening and assuming it's accurate this data is only valid to practice mode but it is valid to practice me it does not scale well to multiplayer when looking at relative performance card to card one good example would be the fury X in multiplayer we ran into frame time consistency issues as a result of memory limitations but those limitations did not emerge in the much more limited practice mode this is a huge difference in play experience even ignoring the higher average FPS the actual frame to frame delivery is significantly different and that's because in practice mode we're not doing with this massive map lots of objects lots of high resolution textures players impact it somewhat but really it's the enormity of the map that really impacts the amount of VRAM utilized and how the card is leveraged elsewhere like in the geometry pipeline for example so the next difference is in the 28 ETI where we see nearly 300 fps and practice but about half of that and multiplayer at 1080p even if scaling were perfect card to card which it isn't really there is still a point at which it's just so far from reality that the numbers are no longer really that useful this scaling is way off the Radeon 7 for example ends up nowhere close to where it should land the 2080 ti does uncharacteristically well versus some of the other Nvidia cards like the 28 or the 1080i the fury X isn't a suffering like it should be and so forth some numbers like the 580 versus the 1060 are actually pretty representative of relative performance in reality so it might be useful there but ultimately this is still somewhat flawed and it's hard to know exactly which cards do and don't perform as you would expect versus multiplayer so we're just going forward only going to use multiplayer testing and our multiplayer test path in the river region as the practice mode is simply not representative enough or load intensive enough for play and cards with different advantages or disadvantages like memory memory bandwidth in the geometry pipeline for example those just aren't leveraged in a way that is intensive enough in practice mode so we hope that apex legends introduces a replay mode for easier testing and assuming it is accurate to play sometimes they aren't and that would make things a lot easier but for now we'll just do it manually in multiplayer mode and practice mode just know that it's kind of okay in some really close scenarios like 1060 and 580 but overall probably shouldn't be using this for benchmarking there's your apex legends performance we didn't go too low and the skew list of video cards didn't really hit the bottom end there but it's fair to say that this this game plays reasonably well on just about everything we tested at we used 1080p some of the devices obviously start to fall off at 1440 you lose a whole lot of them at 4k this is normal but at 1080p with high settings everything we tested for the most part can play it reasonably well and if you did want to use something lower and then we test maybe a 1050 Ti maybe an rx 560 you could still try and do that lower VT Ram capacities do impact performance but you just drop the texture quality so really try and alleviate some of that memory strain and then further of course you could drop to medium or low settings although it starts to suck at that point so everything we test though the the scaling we're pretty confident in with multiplayer benchmarks practice mode we really wanted it to work because it would be the the easiest to do with the greatest sort of reproducibility of the benchmark to avoid any potential variables introduced by multiplayer however after we did run a bunch of practice mode tests and a bunch of multiplayer tests multiplayer is clearly as you've seen the most accurate depiction it is the most reliable for card to card scaling so practice mode although you would think and we did think this that scaling card to card would be probably pretty linear between practice and multiplayer it ended up not being the case and that's just because of how much bigger that map is when you're playing and multiplayer so ultimately we did go with multiplayer mode for permanent testing in the future and just keep that in mind but the game itself is it's reasonably good-looking without being all that load intensive so they're doing some good object calling in there where even though it's a massive map once you actually land in the relevant area you're calling everything of course that's not in the view frustum so the performance is not that bad once you kind of drop in and get into the map so that's it for this one as always subscribe for more go to patreon.com/scishow and exit stops out directly or go to store that gamers nexus net to pick up a GPU artifact in shirt like the one I'm wearing here I'll see you all next time