Gadgetory

so this is a power draw endurance test testing the nrx for 80 on a cheap motherboard and she Paris why we filmed this video before AMD released its 16.7 dot 1 driver update which appears to resolve a lot of the power concerns coming through the PCIe bus and so that's really it's no longer an issue from what it looks like for the consumer because you can pull this driver and basically all concern of over drawn the PCIe bus are pretty much gone from what we've seen still I want to run this video because I am still really curious about the results so we'll play it through keep in mind again it was filmed before the 16th out 7.1 drivers are released or were released so as a user you can pull those down and you won't have this concern really but as a curiosity we're still gonna run the endurance tasks on the RX 480 and without further ado I'll let you see how we're setting up that endurance test for the endurance test we're still going to use 16.6 dot 2 just to see what the original drivers looked like for long-term use all right so we're gonna put a rest to the RX 480 power draw issue my plan here is to use some lower end components to create somewhat of a real-world practical analysis of power draw on the 480 if we're running an RX 480 and a cheapo motherboard with sort of cheap but kind of popular and decent power supply will damage be caused to the cheap motherboard well the PCIe slots suffer any damage or melt or anything like that so that's what we're testing now the issue here very briefly a couple of sites with higher-end electrical testing equipment Tom's Hardware PC perspective some others have been able to test and find that their samples of the our X 480 are not only spiking above the recommended 75 watt PCIe specification which is somewhat normal but they are also maintaining an average higher than that 75 watt recommendation on the spec so it's more than just spiking this is this is a problem because the average is so high on the power draw and it's more of a problem with overclocking but we're gonna test it out so here's what I've got I bought this motherboard this is I think I spent 40 or 50 bucks this on Newegg it is probably a bit lower end then maybe what the average user would pair this card with but we are trying to create somewhat of a of a not worst-case scenario but a bad case scenario this looks like it has a four-phase vrm for the cpu is definitely not a powerful board the power supply I've got the CX 500 I think after Ebates these are commonly twenty or thirty bucks it's a pretty common power spot for a budget bill it's like 50 bucks normally something like that we've got the our X 480 this is a retail model this is the exact same reference card as the press sample the only difference is gigabyte put their name on it but it is a reference card with the gigabyte sticker and that's a that's the 240 dollar one with eight gigabytes and we've sort of tested that already so the test here I want to build this bench in our test platform for 24/7 I will be running a 3d mark stress test on this thing it will burn in and just stress the hell out of the GPU and we're gonna see if it can produce any failures in that period so this will run for a full week 24/7 I'll check on it every 12 hours I'm gonna be doing a lot of logging here actively so first of all before starting the stress test will take some sort of normal use metrics look at the power consumption of the whole system with the power consumption of the card through gpu-z and we're also going to be looking at the thermals pretty heavily so that's something we specialize in I've got this IR thermometer I'm going to use the spot check a few things so we'll spot check the vrm temperatures and things like that on the motherboard spot check its heatsink temperatures and see if any of those electrical components or otherwise begin to sort of destabilize and increase their thermals more than they should over the period of this burn and as a result of the high power drawn a high current high voltage stuff like that this is a thermocouple reader we use this in all of our case reviews and plenty of other things as well it's got what's called a k-type thermocouple on the top of it we can mount two of these to this particular one it has login functions these I will have one login ambience constantly we will use the ambient to subtract the ambient measurement to subtract that value from absolute measurements to create a Delta value this is all standard for our testing the second k-type thermocouple I will be mounting somewhere on the board or the card I'm not sure yet where but I'll detail that when I figure it out as we get into the actual stress test cheap multimeter but this is only to test a few things it's just going to test the power supply we're gonna see if the rails look like they're drawing the power they should outputting the power they should be outputting make sure everything looks stable on the power supply at the end of the test I will recheck it just to see how you know how it's done if it's burned in changed at all things like that in account that in our and our results so that's the test equipment for software I will be using a 264 to log all the temperatures on the whole board all the power draw on the board the video card the CPU everything fans all of that will be logged we use gpu-z perform additional logging on the video card and what else we have here this will not be overclocked if the card and the board survived the first week I'm going to sort of give it a second to cool down analyze everything see if it's performing at the same frame rate as it used to the same power draw same voltage all that stuff will be checked and then I'm going to overclock it and run it for another week and see if at that point we can kill the PCIe slot because of the increased PCIe bus power draw so that's the test that is most of the methodology the rest if there's more that changes will be defined in articles as we get into the actual benchmarking and as far as building this thing let's just look at the the components we got alright so as I'm not grounded right now but it really doesn't matter this is a 55 motherboard or something like that like I said this may be a bit extreme for someone buying an RX 480 I don't think it's that far off of what you would buy a couple with this card but we are trying to have we just create some kind of scenario where a failure is possible if you use a higher-end board especially one like x99 classified that we used for our original testing that's that's going to tolerate the power draw current the voltage of the card a lot better than something cheap that has not invested in good power design on the board so we've got a four phase vrm for the CPU pretty standard two slots I don't even I think this is probably it was probably ddr4 but we have plenty of memory actually a latest ddr4 so we got two slots from memory one PCIe slot that will be used this is x16 if we flip it over you can see it as actually electrically configured for x16 physically so we'll use 16 lanes from the Intel CPU it's an H 110 chipset pretty low-end stuff and I'm hoping I'm hoping that this will create a somewhat representative but not too too extreme use case where we can test the power draw there's an 8 pin EPS 12 volt header despite some of the other boards in this range using 4 pins and then we've got the 24 pin over here of course we're going to keep an eye on the 24 pin and see if any of these traces burn out in this area we've done that before on boards and I'm sure that's where a failure would manifest itself if one were to occur with this card so that's the board the card will of course mount here and we need some memory power supply stuff like that CPU okay so let's choose the CPU of the cooler the memory and all that stuff we've got the power supply and the board already over here so this is some of our stuff I've got it kind of organized for you to see what we have in the shop of parts and first I want to choose an SSD because I already know what I'm going to use these are currently in use I've got some of them labeled for other benches this one is one I bought a while ago and tended to use it for bench never did so we're gonna use that for the SSD and for memory let's go ahead and choose something that's like kind of affordable what is this is this is ddr4 kid of 4 so I've got two more sticks somewhere we can all use two anyway we've got this Ripjaws stuff this is pretty common Ripjaws axe I think is fairly cheap so this is a ddr3 this is 2400 megahertz to 8 gigabyte sticks ddr4 let's use let's use this stuff I think this might be cheaper it's pretty close it doesn't really matter CPU is obviously gonna be Intel as the h1 10 board so I've got a couple CPUs here these are last gen those are 4,000 series these are 6000 series now the thing with any kind of benchmarking test methodology tells you you should only really use one part and test that one card this would be the motherboard in this case but I am trying to do a practical test so as much as I want to use something like a 6700 K with a better power supply with better RAM better everything just to see if we can make the motherboard fail I'm gonna stick with the idea of a semi reasonable reasonable build so we're gonna use this which is an i-5 6400 Intel CPU clocked at 2.7 gigahertz so that's these are our parts right here and assuming everything goes well at this point we can build the rest of the bench so we have to choose a CPU cooler now this is maybe one place where I'll use something higher end than what would be found on a build like this so instead of an air cooler I am going to put a liquid cooler on there that's just because I really want to try and control the thermal values somewhat so we can just focus the testing on one thing in terms of thermals and that'll be the video card and motherboard and hopefully using a liquid cooler will also reduce the chance of damaging a CPU which I don't want to do this is the supply model of one of those popular coolers I know it's used for Corsair you can see it's got no label on it but this is basically a retail cooler that's not been branded yet we've got the thick radiator to 120 millimeter fan with a fat radiator and then a normal CPU style liquid cooling block without the radius sort of machined into it so first things first of course just gonna get the CPU in here and pop this thing out which well hey Natsu now like I said this is probably one of the higher-end components in this build as just because I really don't want to threaten the CPU by torturing it for a week okay okay hi hi but the CPU cooler the radiator right there against one of our high-powered open bench fans video card time okay I really don't like that mounting mechanism but I guess I can't complain it's not a good motherboard the SSD house lied there's rails down here yeah I'll probably put this in the rails and wire it up later power supply smells in there power connects here fan is pointed up that's longer we can breathe so pulls air in down pushes out of course I'm going to turn I'll talk about this in the final methodology video but I need to figure out which of these fans I want to turn on for this test to create a sort of real-world thing so we've got to 120s here we have three up here I'll figure out which ones we're gonna use during the burnin and talked about in the next video one thing here I don't want to I don't want to put this in a normal case because I want to access to be able to probe this thing with like the thermocouple reader an IR thermometer and I want to really be able to see what's happening and be able to interact with it so that is one place we are straying from what you would build in the real world but it shouldn't really matter to be honest so that's the setup that's what we got I'm going to turn this thing on and saw windows wire it up to the power supply and see if we can kill it over the course of a week so check back this will be a multi-part series as we check in on this thing and perform sort of almost like checkups as we go and see if anything's decayed or performance is degrading I'll run FPS tests every 12 to 24 hours and that should pretty much sum it up so thank you for watching if you want to help us create more of this type of content and fund it in the future hit the patreon link the post but the most important thing is to subscribe to the channel and share the content with people because that's that's really what matters is getting it visible and seen so yeah check the link in the description below channel for more information thanks for watching I'll see you all next time