Gadgetory


All Cool Mind-blowing Gadgets You Love in One Place

GTX 1080 Ti BENCHMARKS! Too good to be true?

2017-03-09
what's up guys welcome back to the channel today we are taking a closer look at the GTX 1080 Ti from Nvidia this is their latest GeForce card and apparently their greatest geforce card to have ever hit the market it is supposed to be the single fastest GPU currently on the market and we're going to verify those claims today and see if they're actually true with a bunch of benchmarks where I think we've got like I don't know like 25 benchmark slides or something lined up today so we're going to be taking a very very thorough in-depth look at the performance that can be had when it comes to gaming on the gtx 980ti nvidia was also claiming at their press event week or two ago during GDC that the gtx 980ti would perform up to 35% faster than the gtx 1080 I'm hoping it's true I think we're all wishing for that and more that being said a few specs or a few details about the gtx 980ti if you haven't already heard it's got an MSRP of $700 which has also reduced the price of the GTX 1082 500 u.s. we've also got very similar specs to the Titan X all around we've got the same die GP 102 same chip same number of CUDA cores 3850 for now I did want to specify or correct myself from the unboxing video I did of the TI card I did assume falsely or wrongfully that the base clock was going to be 1.6 gigahertz and boost up to 2,000 megahertz potentially that's actually incorrect the base clock officially in the reviewers guide that I was given from Nvidia puts it at 1400 and 80 megahertz with a GPU boost up to fifteen hundred and eighty-two megahertz and we're actually going to see if we can take that a bit further just by maxing out our power and temperature sliders and taking taking advantage of GPU boost that way as well as of course overclocking the card additionally the ti model only has one less gig of g5x memory than the Titan X Pascal so that puts it at 11 gigs of g5x and this is actually in videos latest iteration of gddr5 X memory which gives it a higher theoretical or effective bandwidth and that's one of the reasons why Nvidia is able to make this a faster card allegedly again from the press-conference than a tight Eck's Pascal unfortunately I don't have a Titan XP on hand so I wasn't unable to test those claims but I will for sure be testing it against the gtx 1080 today we've also got that memory within a 552 bit memory bus interface and also a 7 phase 2x dual FET power design that's capable of supplying up to 250 amps of clean power to the GPU which in theory should deliver a better overclocking experience with less noise and heat we'll also be taking a look at acoustics and thermals in just a bit as a quick refresher from my unboxing video this card does feature a single 6 pin and 8 pin PCIe power plug and we've also got a CVP of 250 watts this is a larger die with it is the GP 102 we're talking about after all which means we're going to be introducing more heat and more power so the fact that Nvidia is also claiming this new new sort of revamped cooler to deliver 5 degrees Celsius cooler temperatures on the die over the GT X 1080 out of the box is pretty impressive and we're going to be testing those claims as well today the last thing I want to say before we get into our testing setup is that all the AI B's the admin board partners will be releasing or slated to release their custom board variants of the gtx 10a v TI later this year in april so stay tuned for those and that's also around the time we should start seeing those overclocked cards from the admin board partners as well in terms of the memory clocks so we're going to be seeing the gtx 1080 shipped with 11 gigabits per second memory of gddr5 x as well as the gtx 1060 which is going to be overclocked to 9 gigabits per second however that card is going to be fitted with the standard g5 x now here's a quick look at our testing setup the notable mention being the 7700 K that we're using and overclocking to 5 gigahertz I wanted to really max out the IPC's so I favored this chip over something like in 1869 hundred K with with a higher core and thread count because we are testing purely gaming at this point and I did want to just make sure that we're running at the fastest operating frequency so that said why don't we go ahead and talk about overclocking really quick now here's a slide that I whipped up that has a couple different metrics including clock speeds temperatures and fan speeds based on three different profiles of the GTX 10 atti which includes its stock right out of the box that's the red bar did such a thing just popped it in and hit go and then also the the black bar indicates maxing out the power and temperature sliders so that we can take advantage of GPU boost the core clock or the memory clocks were completely untouched it was just those power and temp sliders that were maxed out and then the full overclock which is the top gray bar that you see there I dialed in a pretty pretty decent overclock I think I got 160 megahertz on the core clock in terms of awesome offset and then a 350 megahertz offset on the memory clock so you can just kind of see here the red bar in particular for the clock speed the core clock speeds I did want to mention that it wasn't static at 1,750 mega Hertz it did fluctuate in the 1700s I saw it dip into the high 1600s and the low 1800's at some point so it kind of was just jumping all over the place but as soon as I maxed out the power and temp sliders that black bar that's indicating eighteen hundred and ten megahertz that was stable and constant it did not fluctuate from that point and the same goes for the grey bar once we fully overclocked the card as far as temperatures are concerned running stock out of the box the GTX 1080p I was operating at 84 degrees Celsius and maintain that same temperature when we maxed out the power and temperature sliders however if you look at the fan speeds you can see that there was a cost there in terms of noise emissions so we actually did have to ramp the fan by auto fan kicked in went from fifty to sixty percent and that did create a little bit more noise in our in my room now I am testing the card on an open test bench I don't think at 60 percent fan speed you could probably hear the card very well through an enclosed case especially one that had sound dampening material however once we fully overclocked card and release its full potential the max temp does hit 86 degrees Celsius at which point the fan ramps up to 65% and just that extra 5% does make it a lot more noticeable I would say that you probably need a really decent sound dampening case in order to eliminate any sort of noise from coming within your chassis but on that note I'm going to let you guys secure it for yourselves I did do a quick acoustics test so you can hear the differences between each of the three profiles that we're looking at here so there you guys have the sound test and just a quick few mentions before we dive into the benchmarks I was running the 378 66 Nvidia driver for the GTX 1080 and was using a press release driver for the GTX 10a BTI that now has been probably replaced by an official one up on VG the GeForce drivers website but Windows 10 64 bit I was testing I tested all my games at both 1440p and 2160p there was really no point in testing 1920 by 1080 in this instance because the gtx 980ti is anticipated to be so fast that I didn't want to hit or encounter any CPU bottlenecking with those lower resolution tests so on that note slate and gentlemen let's go ahead and kick it out with 3d mark firestrike ultra as our first glimpse is to the performance of the gtx 980ti you can see just stock out of the box just absolutely killer ridiculous victory over the GTX 1080 I'm not going to be saying any percentages in terms of performance games until the very end of all these tests I'm going to be doing an overall average and that'll be exciting so here we have just a really nice score that is a first indication of how we're going to be faring off in these in these tests here so ashes of the singularity at 1440p again we're seeing some pretty nice some some pretty nice traction for the GTX 1080i now even at 1440p in a pretty demanding game like ashes we're still seeing a CPU bottleneck and that explains why we're not seeing any performance gain once we overclock the ti but that being said we're still seeing some really nice performance this game is very playable if you don't mind dipping into the bottom into the high 30s on the very low-end every once in a while with those 0.1% lows kicking up the game before K slightly reduces the amount of CPU bottlenecking here as we see the TI overclocked pull just ahead of the stock TI again just ridiculous frames over the GTX 1080 through and through next up we've got battlefield one at 1440p and no cpu bottlenecking here or at least very minimal results as we see quite a wider gap between TI and overclocked TI 149 frames per second on average with 0.1% lows sustained in the low 90s that's pretty impressive guys and you can just see like going from 108 frames per second with the GT 1082 138 right out of the box we've also got battlefield 1 at 4k again just really good results here 81 frames per second over 58 with GT X 1080 that's of course with the TI overclocked we've got doom at 1440p this game is just absolutely butter with the gtx 1080i whether you're overclocking or not so this is I mean this card is 4k capable I think just right off the bat we can pretty much say that without even having to overclock 0.1% loads over 70 FPS when you're overclocking this thing just an absolute monster when we go over to GTA 5 at 1440p we're seeing another CPU bottleneck here there is a pretty much a stalemate between TI and overclock TI but again we're seeing favorable frame rates when we switch it over to 4k we start to see the game become a bit more GPU bound once again and the 1080i overclocked just pulls way ahead of the GT x 1080 going from 47 frames per second on average to 63 or 66 depending if you're dialing in your own OC hitman 1440p another CPU bottleneck 109 frames per second on average and if we crank the 4k just stellar performance in the average frame rates I think the benchmark just might be a little bit wonky and that's why we got really low 0.1% lows however still very smooth from what I could tell looking at the screen and things like that moving on to Metro last light at 1440p we are bottlenecking the CPU once again it's really no surprise at this point but still getting very nice frame rates in the 100 to 150 range holy moly unless you're talking about the low end which is still well above 80 frames per second Metro last night at 4k yield similarly impressive results this is a very playable game at 2160p you can see 78 frames per second on average with sustained lows in the 50s that goes for the overclocked and the non overclocked Ti and there's the GTX 1080 falling way behind just struggling to hang on at this resolution with frame rates in the sub 60fps on average and we've got overwatch at 1440p of course we're getting ridiculously high frame rates in this game it's not too demanding to begin with but of course this is still very playable at 4k as well it's reduce a little bit and I believe the engine in this game is capping the framerate off at 200 FPS which is probably why we're not seeing a huge jump going from 1440p to 20 160 and also there's probably some elements of CPU bottlenecking I would imagine we've also got watchdogs too at 1440p 82 frames per second on average for our gtx 1080i overclocked with sustain frames in the frame rates in the 60s for our low-end and we've also got the GTX 1080 which is delivering a playable frame rate all around but definitely nothing astounding compared to the GTX 1080p I at 4k the GTX 1080 is no longer a viable solution for gamers looking for a smooth gaming experience we do see the GTX 1080 pull ahead quite significantly although it's not quite the crushing performance victory that we saw with some of our other games tested watchdogs 2 is a very demanding title and especially at this resolution it is playable with the GTX 1080p I more so if you're overclocking of course however it's not quite as ideal as some of the other games that we've run with all those benchmarks run I've calculated any frames per dollar that's achievable with each of our configurations here so we've got the stock GTX 1080 at the bottom there at $500 MSRP delivering twelve hundred and fifty-five frames in total across all of our games today which yields the highest FPD in today's test at two point six seven FPD compare that to the GTX 1080 stock or as I'm sorry 1080p I stock at $700 MSRP with fifteen hundred and twenty five frames rendered that gives us 2.17 FPD so the better value of course would be then to overclock the ti still the same price obviously it's the same MSRP whether or not you're overclocking and at seventeen hundred and four frames rendered you're getting an fpv of two point four three which isn't quite up to snuff with the GT X 1080 it's not quite the value however when you take a look at that gray bar you can see we're rendering so many more frames just across the board that it really depends on what you're looking for you're looking for the best value that you can possibly find at five to seven hundred dollars or are you looking for the maximum amount of gaming performance that's possible within that price range so now really quick before we close this video I did want to give you guys one last look at my D mark which is a very synthetic application where we saw the gtx 980ti pull ahead by 35% exactly over the GTX 1080 in terms of chub factor and that simply represents the likelihood of a particular GPU giving the user a certified chub and this was tested at 4k with tight pants on and stiffy enabled so all-around very impressive marks from the GTX 1080p I overall so now the biggest question that we have to answer still remains is the GTX 1080 TI 35% faster than the GTX 1080 as per in videos claims let's find out if we take a look at our performance slide here and we look at the total number of rendered frames throughout today's testing the middle bar that you see here should be at least 35% faster and 1255 is that the case no it is not according to today's tests and although all the benchmarks that I've run the GTX 1080p I out of the box is 21.5 percent faster than the GTX 1080 which does fall quite short of Nvidia's promise the only way that team green could redeem themselves at this point is that if by overclocking the GTX MBTI you were able to achieve at least 35% faster performance over the GTX 1080 but is that the case yes it is we found thirty five point seven percent more frames were rendered by overclocking the GTX 1080i as found with a stock GTX 1080 and holy smokes guys this is the fastest single GPU on the market hands down I haven't even tested it against the Titan X Pascal but I'm pretty sure it beats it especially if you're overclocking and for $700 I know that's not within everyone's reach it's not even with most people's reach however if you are looking for a 4k experience and you don't want to have to deal with SLI or multi-gpu setup and you've got the money this card is an absolute beast I mean it makes the GTX 1080 just kind of met and the GTX 1080 is a phenomenal card in its own right the GTX 1080i is just an absolute menace of a GPU so Nvidia should be very proud of themselves perhaps I'll be doing a update to hot line in the future hot lines somewhere over there you can't really see her right now but perhaps putting two of those pis in hot line that might have to happen at some point because damn but there's really not much else for me to say at this point guys so let me know what your feedback is I'd love to hear your thoughts on the TI the price point if you think it's worth it what'd you think of the numbers the performance that we've seen today leave all your comments and feedback in India in the comments below obviously and feel free to toss me a like on the video as well if you enjoyed it I spent a lot of time this week benchmarking as you could probably tell from this video so hopefully that's appreciated if you guys like that sort of data then also feel free to let me know that but that's going to do it for now guys go ahead and have yourselves a beautiful day I'm going to go ahead and get out of here edit this video no I got to go to Paul's house and film awesome hardware things are happening all the time things are always happening but hope you guys have yourselves a good day I'll be seeing you guys in the next video and until next time goodbye I was a really abrupt and remember I'm not good with goodbyes they're just so sad I'll see you guys later
We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.