Gadgetory

everyone so we finally finished building our own GT X 1080 hybrid because one was not available at launch so it's a bit of a an interesting project some of it was done and sort of get away but it does work very well and I've got all the numbers for you so I have everything from thermals clock right over time and we've got fan RPMs as well for the erm blower fan so to quickly recap things the past two videos went through the process of me taking this thing apart and then building it as it is now we had some casualties along the way but the original sort of shroud looks like that and that was over here where that EVGA part is now and then this was the original heatsink which we've replaced with the liquid block from EVGA and that's a hybrid cooler which originally shipped on this 980ti hybrid that we took apart for this purpose so this is running nine ATI parts which means that it's not really built for this the cooler itself is fine it's it's the same mounting bracket so everything works just fine there the only thing that doesn't work is this faceplate from EVGA this thing does not fit as we showed in our last video in part two because the reference design for the outer side of the founders Edition card is tessellated in a fashion that these screw holes no longer line up so we could not use that but that's purely aesthetic so it doesn't matter anyway I remounted the original right side of the cooler for the founders Edition just to give the vrm blower fan sort of a pathway to channel it's err that way can I at least cool the DRAM the gddr5 X modules but other than that everything's done under liquid so we've got that all set up for testing so I actually think this looks kind of cool all said and done it's really suboptimal the way it's built because you can't get away with this inside of a case it's just not really gonna work that well we barely got away with it in our open air test bench but I couldn't route the pump the tubes at the top without removing this part of the shroud and I really didn't want to do that because I wanted to direct the erm airflow so you need that for that but it looks not bad really overall considering it's kind of a hack job speaking of hack jobs here's part of the hack job this was I don't know if we even have here's the other half of it this was part of the outside embellishments and it's dead now so oops but let's talk about the numbers we've got the thermals first so equilibrium temperatures are that equilibrium as we call it to make it easy these are the peak load temperatures averaged for the cards that you see right now and the new hybrid the GN hybrid we're calling it runs at 18.66 Celsius delta T for load and 1.95 Celsius delta T for idle that's the GN hybrid then we've got 50 7.51 Celsius delta T load for the founders edition card or the reference cooler that's a massive difference and once you factor in ambient again it gets pretty close to the 80 plus range which is where we saw some throttle and we'll talk about in a moment idle for the founders edition card at seven point nine six Celsius and that is a difference between load and load at liquid versus air difference of a hundred and two percent for the cooling potential for this card so pretty good numbers there let's look at thermals over time this presents the data in a different way as you'll notice the GTX 1080 GN hybrid blows away the competition and it's got a massive difference but both of these temperatures for the founders edition and logy and hybrid are sort of within general operability because we're not overclocked yet so they they do technically both run just fine and we need to look elsewhere for where we would benefit from this crazy Frankenstein project that we've created and you don't really need a benefit it's just fun to do but one of those places is in overclocking stability and as in the previous the review we posted you'll see that the clock rate actually does sort of fluctuate over time with some of the temperature increases and so that's something we explored here as a result of the liquid added on to this card liquid gives us a higher Headroom for OCS higher frequency potential and that's because the 82 Celsius trigger the threshold isn't throttling down the clock rate before we can sustain for a long period that that overclock if that's the major issues is yeah you can create an overclock on the founders Edition that's pretty darn high but it might only be able to hold it for 10 minutes at a time which is great if you're just running benchmarks and you don't endurance test it but if you endurance test these things like we do then it's not going to cut it so we run a one to two hour endurance test which is reasonable really for review purposes and this chart shows the clock rate versus time so this is a pretty interesting one this is overclock the GTX 1080 chokes on its thermals and begins spiralling once it's racked up heat over a period of an hour long test and 82 South Seas is that threshold for thermal reduction for a clock rate reduction on the GPU pursuant to thermals you can see that the clock rate remains stable for good ten plus minutes but starts choking after that the clock rate recovers about 15 minutes later and these dips translate in the real world to severe screen sort of artifacting you'll get either artifacts or flickers or black screens or just frame drops if it's fluctuating up and down really fast which it does in the beginning of the chart it just bounces up and down between frequencies sometimes the range is only 60 70 megahertz but that's still enough to see a reduction in your frame times and that translates to in-game performance let's look at the next chart the first of all the more stable clock rate on the OSI card the hybrido see the card is major and allows us to sustain these higher clock rates but because the CLC keeps the 1080 below 20 Celsius delta T which is 40 something when you factor back in ambient because it keeps it there we're able to run with a fan rpm Auto controlled at around 27 to 30 percent for the VR and blower fan and of course this thing this 120 millimeter fan does produce its own noise but one it's able to spin slower anyway because it's a larger fan so can push more air and - it's it's really not that bad the noise is less obnoxious than a VRAM blower fan fluctuating between 20 something percent and 55 percent which is what you get with the founders edition and we don't get that here it stays perfectly flat at about 27 percent for the blower fan so that's a good thing for noise levels now the new overclock results so originally I was kind of let down because the founders edition card only overclocked two 20 20 megahertz to 20 50 megahertz I had a range of about 30 and that's because the frequency and tolerance from thermals so that's where we were stuck at before and that's not it's really not bad but I was disappointed it so our new overclock we landed at we had a couple numbers here I'm gonna read them off to you maximally we hit 2200 to 2200 to megahertz infer mark which there's a disclaimer there in a second but we had to cheat the voltage away from the memory so in order to hit that core clock I basically reduced the memory clock maximally to negative 500 something megahertz that gave us some extra voltage to pull away for the core clock we're able to sustain it infer mark it did not even come close to being sustained in games so that is a wash that's no good we can't really count that as successful so gamest we're still crashing because the fluctuation was was not tolerable to the GPU and then our final overclock was 2164 megahertz I think that's a plus to seventy-five megahertz offset and the memory OSI was 600 megahertz stable and we were able to actually push 900 megahertz infer mark but not in games so for games 2164 with 600 OC on the memory not as impressive at 2202 but a damn lot better than the 2025 ish to 2050 megahertz we had on the founders edition card so significantly personally more satisfying overclock of course but how does it translate into gaming is the final question here doom at 4k we see sixty two point three three FPS versus 59 FPS 1% 0.1% lows are basically identical that's a five point five percent difference doom 14 41 11 vs 1:09 a one point eight percent difference Mordor at 14 41 16 vs 113 a 2.6 percent difference and Mordor at 4k sixty six point three verse 65 about two percent difference and then finally GTA 5 at 4k sixty six point three versus 65 FPS again about a two percent difference so these are BER clocks in terms of real world gameplay you're not really gaining a whole lot if you're doing this because you want to get more FPS spoiler alert it's like 1 or 2 FPS and it's not worth your time at all but that's not why we did it we did it because the overclock is sustained better over long periods so we can run that higher clock we can even run the the original clock rate we had better because of that so we have less fluctuation in the clock overall which means fewer reductions in frame times or fewer reductions in frame rate as the clock stabilizes for thermals means that the vrm fan runs much slower so it's quieter the GPU itself is way cooler useful for a lot of reasons one is longevity and two is just general heat in the case or in the system you're pushing all that heat out here there's of course heat generation in the and the wires but you push a lot of it out the radiator and so that means it's not getting dumped into your case by another type of fan so those are all good things mostly we just I like to seen the inside of the card and seeing how it builds and how it breaks uh-huh so overall fun projects would do it again would not recommend it if you're trying to get FPS gains if you care about noise and thermals it's a great project to do and but really you should just wait for of course properly supported if liquid cooler so you don't end up with this crazy you know hybrid of 980ti and 1080 that we did I'm happy with it though if you like this stuff comment below we'll try and do more of these in the future pretty fun lab projects build type thing but thank you for watching patreon link in the postal video if you wanna help us out directly with future projects like this and I'll see you all next time