Gadgetory

EVGA is gtx 980ti FTW three is another three fan cooler on the bench today but it's also the most expensive of the 1080 t eyes we've reviewed thus far this card is seven hundred and eighty dollars right now thirty dollars more than the new price of the sp2 and thirty dollars more than the MSRP of the gaming axe and extreme horas card our review of the PCB and vrm for the FTW three gave the card high marks for insane over engineering on all accounts now it's up to the testing to determine if that over engineering offers any functional value for the extra thirty dollars spent before getting to that this in-depth coverage is brought to you by our patreon supporters you can go to patreon.com/scishow nexus if you'd like to help us out directly as that is one of our driving sources of revenue or if you prefer a one-time purchase store that gamers nexus net for one of our shirts our new graph shirts now come in try blends which are very comfortable we've already got a fold PCB and vrm analysis of the FTW three on the channel it's pretty in-depth twenty six minutes long done by residents over conquered build Zoid but three caps on the basics of the cooler and the PCB this is a two slot 1080 TI for some reason that most of the manufacturers this generation are doing two point five or three slot cards for the 1080p is that includes the gaming x and the extreme horas which is the biggest one we've looked at thus far there are very few two slot cards this is one of them as you can see so it is a bit smaller than some of the others we've tested that means it's limited in some ways like the fan size and the cooler depth which will impact performance overall but it's smaller so that's your trade-off now the card is also way over built in the cooling dept this is something that EVGA absolutely lost their minds after ACX and made apparently a resolution to one put thermal pads on everything including the fan controllers and inductors and components that don't need the cooling and to really beef up their cooling design in general so the heatsink alone we kind of detailed in our initial ICS review in February when the icx mine but to recap some of it again they've it's it looks like a standard thin stack so they've got normal kind of thin density and spacing for the fins it's just aluminum fins copper heat pipes throughout the whole thing it's segmented into two partitions for the GPU and the vrm components so that's all pretty normal they've also gone and drilled holes through the cooler through the heatsink so that gives you some more surface area and potentially room for the air to pass through and they've added thin cylindrical fins to the base plate which is different how much does it help as we've said in the past it's kind of hard to say we did a B testing of a CX versus IC x with the IC X review in February and that you should check that out because that has really the the best coverage we've done to date in terms of testing one cooler versus another so that will answer that how much does a health question for this card specifically though you're paying a lot of money for over engineering basically and most of that comes in the form of the PCB rather than just the cooling as always it does kind of start to become a question of how much does this really help you because with Pascal you can over engineer all you want or you can just throw as many fats on as you want you can use this high-quality components as you want pascal still limits you at the end of the day particularly in the voltage budget Department that is an Nvidia limitation so they've gone in and set BIOS limitations to voltage throughput it's something like 1.09 three volts is about your max if you're lucky so at the end of the day you're always going to be limited with these cards but let's recap some of the PCB and BRM stuff and then get into the testing so build Zoids got the full analysis of this on our channel but the basic EVGA is using NCP eight one one five eight DS for the drivers on semi ntp eight one one six two s for the doublers and Alpha Omega semiconductor 69 30s for the memory and core vrm MOSFETs these are dual end sets so you get high side low side and diodes all in one package they are dense the design is a 10 phase vrm but it's doubled so there are five phases doubled because there's no such thing as a 10 phase voltage controller at least not in a side of the industry which helps spread the heat load and power load with a copper vrm place conducting heat to the cooler directly further helping with the heat sinking and dissipation that is part of the over design on the cooler that copper plate is in there it touches through thermal pads to the vrm components the MOSFETs on the doublers directly you can see the imprints on the pad and it's not necessarily needed but it certainly helps drive down the temperatures the review today is going to focus almost entirely on thermals and noise we do have game tests but just to make a point here testing gaming performance with multiple 1080p eyes is going to yield a very boring result we run the test for validation but the differences are going to be limited because it's a 1080 Ti it's the same GPU between all of them and more importantly there are manufacturing variants issues between cards so you could we actually have two of these and they perform slightly differently we have to have a lot of the cards and they perform slightly differently that's because not all chips are created equal particularly with boost 3.0 designs under Nvidia so there's not a whole lot of point to testing gaming on every single 1080i that comes through we do it for validation but this will focus on thermals so keep that in mind so diving into things testing methodology will be linked in the description below as always that's the full article it has all the charts in written format if you prefer the first chart is our clock versus temperature plot which is used under a power virus scenario to draw maximum power through the card and test its clock stability note that this type of test does not enumerate the clock in the same way that a game in workload would the frequency is that lower here than it would be in games but power draw is significantly higher particularly with heavier load on the BRM components the FT w3 is able to leverage it's cooler to maintain a stable clock similar to what we've seen with most 1080 TI cards we've tested thus far other than the F II card a clock has a range and fluctuations of about 50 megahertz one held at 66 C indicating relatively consistent frequency throughout the test that's what we want to see the flatter the better for this line ramping into other thermal tests let's first start with a reminder this is a refresher from our s see to coverage this chart shows the sc2 temperatures of with and without the holes covered in the shroud indicating that the holes in the face plate of EVGA a new design are more for looks and for functionality these are not functionally better for cooling the same holds true with St w3 they're really just there if it looks let's move it to a test we're just out of pure curiosity we unplugged the third fan and ran the other two at fixed speeds this test was conducted with all fans at the same speed the only difference was disabling the power fan on the second test just because we figured the over-engineering on the BRM and the cooler would be enough to keep the power components within spec without that third fan the result is mostly what you'd expect running two fans that results in higher temperatures but despite being somewhat expected this data is still really interesting we don't normally have this level of access to temperature sensors without placing them on the board ourselves and often them of that to two to three thermocouples because there are nine diodes on the IC X cards we're able to fully understand the impact of a2 fan versus three fan cooler across the PCB not just from the GPU and our own thermocouples so using EVGA NTC thermistors as far as power component temperatures are concerned it's really not that much of a loss to move to to fence the overbuilt vrm and the heatsink through their job keeping the vrm components within spec in our tests even that 87 c number for power 4 isn't terrible but power 4 isn't a MOSFET on this card it's actually the back of the PCB behind one of the RAM components of the GPU but still not that bad what's more interesting is that the third fan which EVGA has positioned over those components seen it's the most heavily impact the GPU temperature itself GPUs diode temperatures increased by roughly 10 C without the third fan spinning despite the fact that the GPU and memory fans the left-most two fans are spinning at the same speed as with three fans this indicates that the heat pipes and right side of the sink are dissipating a significant amount of heat from the GPU silicon and that the third band is necessary to dump that heat from the fins we can validate these numbers by looking at GPU backside temperatures which have risen about 60 also the memory components really don't seem to care either way they are certainly benefitted by the third band the difference is about six or seven Celsius and those memory modules are rated for well over the heat that they're experiencing in both tests so although that's not something you're likely to do disabling the third fan it does help us understand how the cooler functions the third fan is not really all that important for power temperatures because they're still way within spec which means you can build a fan profile to spin that fan slower and not have to worry about your vrm components like the feds heating up too much so that helps with noise when you're profiling the fan but GPU temperature does go up a bit because you're losing some of the dissipation potential over this entire half of the heatsink which obviously touches a bit of everything on the car not just power components this next chart shows temperatures versus the sc2 card unfortunately half of this chart is useless here's why EVGA positions its sc2 and FTW three thermistors in different locations on the board which we can show on the screen now this means that despite really wanting to compare the two directly we really can't power reading for for instance is really warm on the FTW three a much warmer than on the sc2 in the side-by-side comparison but it's also on the back of the PCB with FTW three located behind a VRAM module and above the GPU backside that's going to be a really hot spot of the board and the sc2 positions its power for module on the front side and within the mosfet region so it's measuring a completely different location on the board and even a different side going back to our thermal chart between the two you can see how this difference plays out power 4 on the sc2 it looks significantly lower but it's also measuring a different component these are therefore not directly comparable aside from the to GPU temperatures and some of the memory temperatures fortunately we also position our own thermocouples on all the cards and put those in the same spot each time so we can directly compare these readings let's look at the s TW 3 components we measured with thermocouples as opposed to the IC x NTC thermistors then compare those against MSI gigabyte and the NVIDIA F II cards we're still expanding on this chart this is brand new and we're excited with more detail for it in the future including V RAM and a fixed 40 DBA output test for now this is where it's starting with auto settings for the coolers all cards running out of box clocks we see the EVGA 1080 w3 keeps lower-powered temperatures than everyone else meaning that they could sacrifice some of the cooling department for lower noise this is a common trend with EVGA where the auto fan is aggressively profiled compared to the competition at 63 up 7 Celsius for the power measurement on the middle of the Fed's EVGA runs at nearly 20 see cooler power components in gigabytes whoo-whee remarks ran warm in our extreme RS review and about 7c coolers on the gaming X so far the EVGA FTW 3 is the coolest 1080 Ti that we've tested to date for both the GPU diode and power component temperatures but there's another component to a cooler and that's noise until we add our 40 DB a fixed temperature testing let's look at the normal noise charts under idle conditions the fans spin down to 0 rpm just like other cards do right now outside we've noticed that the fan does like to spend up to 700 somewhat regularly on the GPU but you could shut that functionality down with the custom curve Auto conditions place the EVGA FTW 3 at 42 to DBA whereas the msi gaming addict operates about 36.5 DBA and the gigabyte extreme horas is about 38.8 DBA considering both of these cards running warmer than the FTW 3 which is the coolest of the 1080i is thus far it makes sense that the quieter cards would run a bit warmer the really interesting test will be our fixed DBA temperature tests which are still underway and we posted separately regardless for now the EVGA FTW 3 sits perceptively louder than the extreme orison gaming XO the gaming X deserves praise for its significantly lower noise levels which are made possible by the fact cooler and taller high defend EB J's FTW 3 fans are about 90 millimeters across whereas the gaming ICS is about 100 millimeter fans let's move on to some brief game tests our overclocked stepping charges on the screen now we weren't able to push our FTW 3 is too high but that's been true for basically every 1080 I would come across thus far these cards are already up against frequency limits on Pascal and aren't particularly impressive overclockers with thanks to voltage limitations imposed by Nvidia's bios running ghost recon at 4k the 1080i FTW 3 is the fastest of the non overclocked 1080 TI cards running an average FPS only marginally higher than sc2 and this is within test variance by the way so we can fairly state as the SC 2 and have TW 3 are functional equal in average FPS gaming X is next down the line of 58 average with the extreme or s at about 57 average there's no meaningful difference between any of these cards when it comes to pure FPS which makes sense in a world where boost 3.0 it largely equalizes cards at the same chip anyway overclocking that tw3 isn't particularly exciting either despite the PCBs higher quality will highlight those numbers anyway but again it doesn't really matter how much you engineer these boards without power mods and hard mods you're not going to be able to overcome the voltage and power limits established by Nvidia and Nvidia establishes those they say for safety because the 400 series and 600 series had a lot of people over volt in beyond where they should have at 1440p Ghost Recon performance is largely the same stack FTW three runs an average of 95 with 1% loads at 82 and 0.1% loads at 80 comparatively the gaming X operates at 92.7 average with the extreme Oris also at 93.7 FPS average and the sc2 at about 95 FPS average overclocking helps out marginally once again but we don't quite break 100 FPS running doom with woken at 4k that uga 1080i FTW three is once again the fastest stock card on the bench even versus the Titan XP in this case and it outperforms the our gaming X by about 1.2 percent that's certainly not jaw-dropping gains and can be partially accounted for with chip to chip manufacturing variants but it is a lead nonetheless the FTW three also outranks our gigabyte extreme card by 4.5 percent as the extreme Oris is running into power limits under stock conditions without pushing the power budget further overclocking moves ranks around a bit but again games are limited overall for the 1080i class cards Sniper Elite 4 with DX 12 and a sync compute lands at the 1080i FTW 3 again at the top of the stacks of 1080i listings just barely ahead of the Amazon X in fact these two are effectively identical as they're within test as variants where to F gets ahead of the extreme card or about 2.8 percent which does fall out of variance but it is not really that exciting if you really want more game benchmarks they will be in the article links in the description below it's just that we're looking at something where thermals and noise matter a whole lot more than gaming performance because once you've established a baseline for gaming between the AIB partner cards it doesn't change a whole lot it's not like they make the actual Jeep you that's making the cooler and the bored so if you can run into limits no matter what so the gaming axes and extremes can swap places between reviewers just because you might have different quality chips and the same is true here this is on our bench the highest performing stock card but at times that's a difference of 1% or less in some cases so it's not FPS thing we're talking out here in terms of thermals and noise which is what actually matters this runs a more aggressive fan profile than some of the other cards the gaming acts included so that means it's a bit louder but it's also cooler just one of those trade-offs you could build a custom fan profile with this and probably should if you buy one because that power fan doesn't need to run as fast really none of them need to run as fast as they do you could run a higher temperatures with a lower noise level probably get around or under 40 DBA and still be just fine in the temperature Department unless you're in a really hot ambient environment so that stated just like the SE 2 you should tune fans if you buy this card but this is a $780 card so it's a tough sell the others are 750 the FC 2 is originally something like 720 which was a really good buy now at 750 I think there was a pre-order thing but at 780 we're 30 over every other card we've reviewed the PCB is fantastic engineering it's really well done and builds or breaks it all down in our video the cooler is good engineering the icx stuff is pretty cool but it does not make the gaming experience better that's just it's how these things work so what you're looking at is short of doing shunt mods and hard mods you're stuck in a power limit and a voltage limit that means you can overclock more than the competition in fact most of the overclocking limits are going to be more chip the chip variance from the PCB itself unless you buy a really really bad PCB so it's it's just a question of what is what do you get for an extra $30 over and I see two answers a fan so what do you do with that okay you can run them at slower speeds and get lower noise output maybe decibel or two so maybe that's valuable to you but the FC 2 is already pretty good it's $30 cheaper it does all the same gaming overclock and everything really a bit quieter and is reasonable in every aspect as well extreme Morris is okay it's the warmest and it's louder than the gaming acts not really I'd be looking at the gaming acts sp2 or the FCW three and only the FTW three if you really really want it for some specific reason otherwise the sc2 is a better buy at the price both of them are two slot coolers which is perhaps the primary difference an advantage over something like the gaming X otherwise back to what we were saying on the previous reviews at $750 there are a lot of cards that are 720 and 730 from EVGA included that would probably suit you just as well but have slightly less flashy coolers and components and still perform more or less the same noise levels on these seven $50 cards though are pretty good especially if you're going to custom tune them so that's it great PCB good cooler it's a little louder than the competition a little cooler as well so just comes down to how much you value those things but I'm having a hard time seeing why to buy this over an SC to considering for 99% of people they are functionally the same so if you know a reason great buy it for that reason but I don't have one for you so that's all as always patreon.com slash Khmers and access that without directly you can hit the link in the description below for the full review of this card with all the charts in written format or store it out gamers nexus dotnet to pick one of our shirts we have the GN grab shirts now in tri-blend so definitely check those out thanks for watching subscribe for more I'll see you all next time you