Gadgetory

what's up guys Jays $0.02 here and I just spent all day testing all these graphics cards from the 10:15 on TI all the way up through the new Titan X Pascal hitting pretty much every GPU in the stack to find out whether or not bottlenecking is still a thing in 2017 JC sense is sponsored by precision camera and video Texas largest photographer and video retailer located in Austin for over 40 years they're giant 20,000 square foot showroom floor local workshops and worldwide shipping are just a few reasons why you should head right now to precision - camera comm the reason why we're even talking about this is because any time I do a video about lower end CPUs the conversation of bottlenecking always comes up and that's something that we haven't heard as much I don't think as we moved forward in the in technology since my last video which was 2013 where I should use my 3770k for that one if you talk about timing on that anyway I don't think it's become as commonplace of a discussion because CPU technology seemed to move forward at a slower pace than GPU technology which means that in my mind actually we could have probably seen more likelihood of bottlenecking as the CPUs slowly progressed and GPUs got balls to the walls fast compared to anything available back in 2013 like even the 1050 which you get for your money out of 1050 is way more today than what you would get for a hundred bucks or 120 bucks whatever it is back in 2013 now if you're new to pcs and you have no idea what bottlenecking is it's basically what it sounds like it's when your time trying to take something really big and cram it through something really small I'll let your imagination fill in the gaps when you have a lot of performance potential in your graphics card speaking to your CPU but the CPU can't handle the information as fast as it needs to to be able to render all of those frames on your monitor so what you never ibly get is a GPU that's sitting around waiting for the CPU to process the data that's being sent to it and then you start to get dips and Peaks and and stutters and slowdowns in your gaming experience now in a nutshell that's what bottlenecking is but it can be much more detailed and much more complicated than that depending on where your bottlenecking why your bottlenecking it comes all the way down to what CPU you're running obviously the GPUs and more Courtley the titles that you're playing because no matter how I did this test someone's going to complain that I did it wrong because there's no way to test for every single permeation possible when it comes to hardware and software integration of the way that they work with each other it might not look like it's bottlenecking much on one title then you put in another title and it's just an entirely different result so for the sake of simplicity today I just used superposition we're using the 1080p medium because the idea is by not cranking the settings we're offloading as much of the stress onto the CPU as possible because the GPU is just running away with performance and it's the CPUs job to handle it so that's how we're trying to introduce bottlenecking and our tests here but like I said I tested with the 1050 1050 TI 1060 1070 1080 1080 Ti and the new Titan X Pascal the the newer new one because there's a couple of them it's the newest one that's all the matters but I did it with the G 45 60 Pentium CPU which is like an $80.00 CPU and then I also did it with my I 5 7600 on a z2 70 motherboard the Pentium was using the h1 10m like we used in the budget build so we've got two completely different systems there in terms of spec what the same exact GPU is being used on both systems to try and determine where the bottlenecking actually starts now before we talk about the results I'm gonna go ahead and show you those right now I'll be the first to admit that my predictions were actually inaccurate I I thought we were gonna start seeing bottlenecking sooner somewhere around the gtx 1070 where the 1070 would be the last GPU that we saw performance matching and then the 1080 was where we start to see some deviation in performance but I was wrong the 1080 on the G 45 60 gave us the exact same score an exact same fps as the i5 7600 K it was wasn't until we went to the 1080 Ti that we saw things start to run away now that's actually a good thing because that shows us that even a modern low-end like entry-level CPU is able to handle all the way up to something like a GTX 1080 which is the tenth if we compare the performance Delta between the 1080 today and what was the 80 series in 2013 which was what the 680 my transitioning to the 780 there's so much more performance today to where that means that you could build a pretty good system out of something like the Pentium chip as long as the title again you're playing are not extremely CPU dependent because what's gonna happen in real titles and not synthetic benchmarks is you're gonna see a lot of things happening where the CPU is having to load textures and it's having to load maps and other players and game data and then you'll start to see your performance change quite a bit obviously depending on what the CB is having to do besides just the FPS the GPU is running at it so that's why I say that there's no test I could have done today that would have been you know appeased all of the crowd but what I will do today as I will do this I've actually got the i-5 7600 K right now set to two cores it's a two core CPU there's no hyper threading on the i-5 right so it's a two core CPU that means there's two less threads then we saw with the Pentium chip and I'm gonna go ahead and run superposition now and we still have the Titan XP installed right here and we're gonna see what kind of a performance difference we get when we take two whole cores out of the equation so technically now we have two less threads in the Pentium chip so I'm gonna let this run and we'll compare the score to see how much it actually affected something like the i-5 now the test is get ready to finish right here you're gonna notice quite a bit of up and down in terms of GPU usage when we look at MSI Afterburner but the score is undoubtedly going to be different I mean we introduced bottleneck so we got a twenty thousand three hundred fifty five when previously we got twenty 1993 so we drop sixteen hundred points and our max if our average FPS went from one hundred and sixty-four 0.5 FPS to one hundred fifty two point two five but that's still super impressive to me because it shows what the IP sees four instructions per clock are actually able to achieve on modern CPUs back in the day if I if I did this test with an i5 from 2013 and disabled half of it I'm not even sure the test would have run I mean it's possible wouldn't even run now something else I'm gonna do here is I'm gonna show you why even though I'm synthetics you know it you can see the variance here why in games it's more than just about how much of bottlenecking your GPU it's about how much you're just vital locking the entire processable processable processing capability of your system because of the fact that you have to account for more than just your GPU right the CPU still has to run the game it still has to handle everything happening in the game on top of the visual process that the GPU is adding to the experience so if we go into like rise of the Tomb Raider here and even on DX 12 which Det API is designed specifically to try and alleviate some of the overhead with asynchronous compute on when it comes to games and allow lower SPECT Hardware to be able to have a better gaming experience I mean we Vic's we've really stretched out the capabilities of DX 12 here right two cores no hyper threading and it's heightened XP I really hope nobody's running this config out there but what you're gonna see right now the reason I'm still talking as I'm buying time to show you what the load is like because we're still running a Samsung 850 Evo SSD so read/write is not our bottleneck here either cuz that's another possible bottleneck right hard drive versus SSD read/write makes your computer feel slow because the CPU is waiting for your hard drive to send the information it's the same thing just a different perspective where the CPU would be normally waiting for your GPU but when we were still loading we're still loading right now this is crazy so now if I say start benchmark check this out in fact I might even just have to stop watch this or something do a jump cut because it's gonna take a while now to get through this test it's gonna take us a bit so it just started okay and the FPS is actually not that bad but look at the GPU usage 62% 65 64 that is a bottleneck because now that the CPUs have in hand handle other things besides just the GPU like in the synthetic benchmark you can see here that the GPU is not being fully utilized the busiest way to determine whether or not your bottlenecking your GPU is to load a free program like msi afterburner put on the OSD or the on screen display and then you can actually monitor what's happening here in this number specifically is our GPU and now that we're loading screens here this is the part that takes a little while so I'm gonna go ahead and start my stopwatch so that was fifteen point six seconds plus the roughly the ten seconds before I got this going so about 25 seconds between scenes as the kid goes by but this keyboard is that loud here yeah there's new mic again you Mike it's good but bad it's good because it sounds great it's bad because it picks up at every your CPU is more important than just how much Fe FPS you're losing from your graphics card because of bottlenecking idle necking is a big picture thing and that's why I really hate that the way that the bottlenecking terminology is thrown around specifically and you're gonna bottlenecking that GPU with that sheep you it's like you're gonna be really just bottlenecking your entire experience where many times I could say bottleneck e in this video before people lose their I hate that term I hate bottlenecking come on come on a few moments later oh there we go so it just started and it's like two minutes in like 49 seconds or 50 seconds it took me a couple seconds topic but almost three minutes to load that transition the crazy part is all the sound already played and now the video is going without the sound they're like completely unsynched itself unsynched and it's a in sync that was a terrible band in the 90s our group won't calm a band any play instruments all right so our overall score was a hundred and six point eight two fps so what I'm gonna do now guys I'm gonna go into the BIOS I'm gonna turn that those two cores back on and run the test again so one hundred and six point eight two can you remember that all right so active cores are now set to all and now we can get moving in do you see the difference in FPS though if you look at the other way it was that like 70 FPS a FPS look at the F it if you guys were wondering how strong that Titan XP is this not even overclocked 200 as a stupid it's stupid but more importantly they look at the GPU usage now that we're sitting up at what 98% 99% and that's wrist where it's gonna stay it's just where our are supposed to see that I mean how fast it loads but this is an issue that we were probably below the specs required by this game I believe a quad core is recommended and required for this game it's like I mean two two cores it's just it's cool and all it still did well once the scene loaded and stuff but yeah I mean you see how much faster that went before that that's the one that took two almost three minutes to load it was like what ten seconds all right we go so one hundred and eighty four point nine one FPS versus the previous test which was a one hundred and six point eight two as I believe it was so there was your real-world scenario tests with bottlenecking again I know folks will be like lads synthetic it's not it's on a rail but it's still using the game engine whatever so they'd the real takeaway from today's video though is that modern CPU is even the budget CPUs are still pretty impressive on how far we could actually get down the stack before we introduced any real slow down and bottlenecking the 1080 TI was where it was so somewhere really between the 1080 and 1080 TI like if there was a gap even but like a card in between there that's probably the cusp of where we were starting to see the problem but yeah I mean I don't think anyone's gonna run a pentium CPU with a Titan X or a Titan 1080i anyway even a 1080 no it's not gonna happen I think something like a ten sixty maybe a ten seventy so if you're gonna be if you're worried about bottlenecking just don't go completely mismatched on your hardware and you'll be perfectly fine same goes for like the AMD stuff maybe not so much the Athlon though but again that's a that's from a core back in that 2012 2013 era which is why we saw some of those problems but anyway guys I'm gonna go thanks for watching today's video these topics are always fun and remember they're requested by you guys so go and check it out also to remember you can buy merch not this merch III this isn't my store right this is this is Kyle's I wear this shirt even though it's too big for me it's like a tuck I'm wearing a blanket because he hates that with a shirt he tells me to burn it every time he sees me wear it he's still awesomesauce as far as I'm concerned he you know bit wit your awesome sauce thanks for watching guys and I'll see you in the next one you