How to stress test a PC to find errors and crashes

[Music] all right guys we got a somewhat quick and easy one for you today i'm going to try and keep it short and concise uh we've talked about how to build towers for years on this channel 10 10 years now we did a video that's actually gone quite viral and gotten over 10 million views now which is what to do after you build your tower which is how to get your drivers installed your operating system uh chipset drivers bios software installed all that sort of stuff but we're going to continue on now from that which is now showing you how to stress test slash test and make sure all your components are working exactly as they should nzxt's build is a quick and easy way to get a new gaming computer and right now they're proud to announce expansion and availability to australia the netherlands france and italy build a gaming pc on your budget using the built-in configurator and see exactly how your favorite games will perform want to build your own pc but still have the nzxt peace of mind warranty then the new bld build it yourself kit has what you want buy it and build it yourself and nzxt has you covered to get started configuring or building your next gaming pc visit the build link in the description below all right so we're again we're assuming that you've already built your computer and you've already followed our tutorial on how to get your os installed and all the software and stuff ready to go so when it comes to stress testing your stuff a lot of people think you only have to do stress tests if you have overclock supplied or something custom but that's not true this is actually a really good way to test to make sure your thermal paste is properly applied because your cooler can often the temperature look totally normal as you're operating the computer and browsing around and stuff but you start gaming or you start maybe trying to live stream put a little bit of stress on the cpu and suddenly start getting the blue screens or the crash to desktops or the hard locks where your mouse stops moving and the only way you can get your system back is to flip the power switch on the back of the power supply and get the system to come back a lot of these types of issues could be a due to excessive heat where you may be your thermal paste or you forgot to put the thermal paste on entirely believe it or not your system will operate normally with the heatsink touching the top of the ihs and the cpu without thermal paste until it's under load then you'll notice things start to act really weird so it's just good practice and good rule of thumb to just stress test everything under unrealistic scenarios like for instance cinebench now when it comes to testing cpus it doesn't matter if it's intel it doesn't matter if it's amd cinebench is a test we like to run because it gives us the worst case scenario when it comes to heat so that's why the very first thing i will do is i will always go to cinebench and start running a test on that which is a loop because we want that heat to generate two things to keep in mind before we do this one if you're running an air cooler it will take a very short amount of time for the cpu to reach its maximum temperature equilibrium equilibrium means where the cooler is fully saturated it is working 100 of its capabilities water on the other hand whether it be a custom loop or an aio takes much longer for the thermal capacity of the liquid to be maximized which will then give you your equilibrium typically it's about 30 minutes before you will actually max out your temperatures on aios custom water loops can take even longer if you have a large capacity so don't put the you know a two minute test together and say hey my temps look great and then suddenly you go and you start doing long rendering sessions or long live encoding sessions and then suddenly you start noticing weird things happen because if something is less than you know desirable on your system and it starts to heat up over time and it continues to heat up and then you start getting into that throttle territory and things start slowing down that could be because something's up with your cooling system and you didn't test it long enough to actually uh expose that particular problem but you need a way to monitor your temperatures and your cork locks now we're using ryzen which means that we can actually use ryzen master to keep an eye on our temperatures and our clocks i do have pbo or precision boost overdrive activated on this cpu precision boost overdrive is an automatic overclock of the processor now ryzen master allows us to see right here our temperature the top number that's fluctuating is our current temp the number at the bottom is going to be whatever the tj max or the thermal junction max of your cpu is that is the hottest point of the cpu how loud how loud yeah loud technically because fans and stuff how hot it's allowed to go before it starts to slow down the clocks to control its temperatures if the temperatures get out of control the clocks continue to go lower and lower and lower and lower until it finally determines hey we are completely overheating we're in meltdown mode where the system will then shut down if you start to see and get anywhere near 95c then you definitely know that you have a problem with your cooler your fan step your thermal paste something even a stock wraith cooler or something on here would be able to take a pbo or precision boost overdrive and still keep it underneath it's tj maxx we're using a dark rock 4 here which is not going to get us anywhere near that because it's a much bigger cooler than we need with this cpu now the other thing too is you will be able to see right here the peak speed this is obviously our core clocks right here um right down here underneath where it shows the the ccx or ccd0 this is our current speed if you're running an intel cpu you'll want to download a utility called xt or extreme tuning utility that is an actual piece of intel software that will communicate directly with your cpu it will show you your current clock speeds it'll show you your status of the cpu is it throttling everything you see here in ryzen master is going to be available also on intel's xtu so you'll want to take the logic we're applying to our testing and just use xtu for intel horizon master for amd the other thing we want to test here is what is our power setting set to now we are out of the box dock settings here so i just went to start i typed power went to power settings go to additional power settings we are set to balance down here under additional would be high performance i'm leaving it balanced this is a gaming machine we don't need to run full speed all the time we want it to fluctuate that way temps stay down power draw stays down until it's needed then everything comes back up so you can set your testing here 10 minutes test for throttling 30 minutes to test for stability so it even tells you here now cinebench is only one utility it's only one type of instruction it's only one program and it's not going to give you the full indicator of the stability of your system i've found though typically if we can run cinebench without a problem for the 30 minute test we can run pretty much anything without a problem i find this to be the most demanding test but there have been times we've gotten overclocks to be stable in cinebench and then suddenly going to premiere and funky things start happening has everything to do with the way the cpu is accessed and the other thing too is when cinebench fully loads the cpu the cpu's clock speeds will kind of drop a little bit because of power limits and all that sort of stuff which brings the clock speeds down whereas other programs that are not as demanding might allow the clock speed to stay up a little higher where it kind of fluctuates up there and some of those fluctuations between say five gigs five two five one five two where you start to notice instabilities whereas when it drops down to like four eight or four nine all core and like an intel cpu they're actually more stable because the clock speeds aren't as high so you're gonna want to really test this out in your your programs you're expecting to use the cpu in before you start doing critical tasks like actual work or rendering for jobs or whatever before you start to say okay i'm fully stable you might lose a lot of time with work uh by losing data and stuff because of crashes so with that out of the way make sure you have your system set or your test set to your minimum duration 10 minutes is fine for an air cooler and we're going to run multi-core because single core is only going to give us the single core performance multi-core is going to show us the stability so once you hit start it's going to take a minute for it to start going and you'll notice the temperature starts climbing on the cpu before the test ever starts running but as we can see right now we're running at 4.625 gigahertz we're only hitting 6970c so if we do the math 95 minus 70 we've got 25 degrees celsius of headroom right now before we would even start to throttle but what we're looking for here is our temperatures and our clock speeds because our temperatures are so low and so far below our limit we are maintaining our full boost clocks with our pbo precision boost overdrive what we want to see is if this number starts to drop we immediately would want to look at the temperatures typically that number comes down if the temps are too high if you have an overclock enabled which is lifting all those power limits but already i can tell you right now with an air cooler what the temps currently are and you can see they're not really climbing we've gone up what not even 2c since we started the test you'll notice these temps now i could let it run for a half hour and probably never go any higher than 73. that's typically what you'll find with an air cooler like i said water cooling let it run for the full 10 minutes let it go for 30 if you have the time that will give you the true indicator of your test all right so i'm pretty convinced that we're fully stable here our temperatures are still at 73.2 73.4 71.8 so our temps are fine our clocks exactly where they're supposed to be 4.6 gigahertz i just want to go ahead and stop this test right now because i want to show you what it would look like in a stock situation if we did not have precision boost overdrive now if i run this same test we're going to notice two things the temps are going to be much much lower but look at our ppt it's maxed out this is saying we are using a hundred percent of these 76 watts available to us our clock speeds have dropped all the way down to 4.15 gigahertz all core that's specifically because of the power limit so this is one of the reasons why enable precision boost overdrive on amd cpus in fact i can even hit apply on that while we're running the test and now you can see it jumped right back up to 4.6 back up to the 71 to 73c and then all these power limits have been lifted they're just an arbitrary number there has to be a limit unlimited is not a a mathematical factor so it has to put a number in there so could you imagine a 540 amp draw cpu so what i would do next if i was testing this for full stability is i would uh take some sort of a project a premiere project or something and i would create a an encoding timeline that's designed to be stressful lots of transparencies lots of overlays lots of transitions and then maybe take the timeline and duplicate it a bunch to where i'm trying to make the render take a long time 30 40 minute render then i would do a premiere render and see if it passes and maybe do that a couple of times if it does then typically we're good enough to call our cpu ready to work let's go ahead and call this cpu good and let's move on to the gpu the gpu is the other thing where if you're applying overclocks um which we're not in this instance uh you would start to notice issues nearly right away when it comes to your gpu so i'm going to use heaven for this but for this it's a benchmark but it's a looping benchmark now we're going to be running this in tessellation extreme anti-aliasing 8x we're not going to go full screen we're going to go 2560 by 1440 even this is a 4k panel and then we are going to be running our preset custom because of everything we just added which is custom we are running amd radeon software if you were going to be using an nvidia graphics card geforce experience has telemetry built into it now so you can see the temperatures and the core clocks and all that sort of stuff and we are going to use the default overclock gpu that's a one click one clock it's a one bar overclock so that you can basically allow the logic of the graphics card to overclock itself slightly i'm going to start there i want to see what it does that's why i run this in non-full screen i run into windowed mode so it will still go full speed in the background but i can have an overlay going right here but our current clock speed went up to 26.65 yes amd does have a superior clock speed to nvidia however nvidia still has a superior instructions per clock than amd so it's kind of funny how that balances out all right so amd smart ass memory is enabled by default because it's named motherboard about the latest bios with smart access memory it's on by default um but what i'm looking for here a couple things clock speed i'm looking for major fluctuations will that start to drop if i started to drop then we're going to see problems probably with thermals on our gpu i'm also looking for what our current temp and our junction temperature are now current temp and junction temp these are two very different things basically current temp can be looked at as an edge temp or the lower temperature cores of the gpu now there's temperature dies all over the gpu memory temp drm temp edge temp junction temp memory temp there's all sorts of temperatures but what this is giving us here is our range our coolest temperature reading on the gpu die and then the junction temperature is our hottest so we're looking for here is what is that spread now that spread on this this particular gpu is fairly low our current temp is 50c our hottest core or hottest spot on the die is running at 65c that is really cool absolutely no issue whatsoever um it drops in between scenes as you can see right there we're from 2666 down to like 10 something and it jumps back up heaven in between scenes reloads another scene so typically if you're gonna find a crash with heaven it's gonna happen in between the scenes where the screen goes black and as the scenes the new scene comes on then you'll see a crash because that's where you get that sudden spike in frequency and it's those spikes that are going to basically expose any sort of a stability issue in your system it's not necessarily the constant load that causes a crash it's when the load is lifted and then suddenly the megahertz jump up real quick that's typically where you'll find uh any sort of an issue there but what i'll do now is i'll let a test like this run for shoot we've let when we do some of our stability testing here with overclock we've let it run for hours but what i will do once i determine heaven has stress tested it enough to determine its heaven stable then you can move on to something else whether it be a game like i said 3dmark 3dmark's not free but 3dmark does have a stability test built in we'll take a 3dmark test and it will loop it if you pass that you can pretty much always guarantee you're going to be stable out of the box stability test very rarely yields any sort of instability problem for the with the the stuff usually if any sort of crash happens in this scenario we just test it within means we either have a bad installation of our cooler for the cpu or the cpu slash gpu themselves have some sort of a defect in them and if you start finding where stock settings are crashing that's an indicator something is bad with the product itself whether it be they forgot to put thermal pads maybe on a vrm or thermal pads on the memory or bad application of gpu paste yeah we've seen it before we've seen bad applications gpu paste where a whole piece of the dye is not touched at all by thermal paste that's the kind of stuff that can indicate and be shown by these basic tests that we're doing all right so the last we're going to test here is our memory you'd be surprised how many most of the time people message me and say hey i'm getting this really weird issue and these crashes and stuff um is there something wrong with my memory you can test that there's actually free utility called memtest x86 and what this is it's an it's a software that will run on an external uh source or a usb stick so i'm not going to show you guys how to install it and so if you guys can go and find videos on that it's pretty simple so i'm going to go ahead and put in the memory stick we are going to reboot to this drive and then it will give us uh it will start our mem test and what that's going to do is it's going to test every single sector and chip on that memory now to get into your memory or your usb stick uh every single motherboard is a little bit different on how the boot overrides work so you're gonna have to revert refer to your bios and your particular motherboard to determine the best way to get into that bootable media you might go in and change your priorities you might have to go into bios and then do a one-time boot override in the nzxt board it's hit f11 for us it's a corsair voyager that's 32 gig stick right there partition one now it's booting up mem test and what's going to happen is it will automatically once it tests everything it's going what is your ram what's your ram's controller oh this is what you've got this is what's expected to work and this is how it's expected to run so it's going to start automatically in six seconds once it starts it's now checking every single bit and byte and but they're all the bees of that those memory sticks individually to determine if everything's running properly if this passes we can pretty much guarantee that our memory is fine now i am running the xmp profile or the docp remember for amd we are running the extreme memory profiles remember memory by default does not run at the advertised speed you have to go into your motherboard and you have to enable either the xmp or the docp to get the advertised speeds of your stick i get people all the time i even got one yesterday something's wrong with my memory it's supposed to be running 3600 it's only running 2133 help and all these these exclamation points and it's like i talk about this every time we build a system but it still surprises me that the the memory manufacturers don't put even anywhere listed on their memory like put a slip put some sort of slip of paper that says warning must enable these speeds for these speeds like they don't tell you anywhere that that's the way that it works they just expect people to know and unfortunately people who are doing this for the first second or even fifth time don't always know uh but anyway it's going to tell you what it's doing on the different tests so right now we're in test six block move 64 byte block so it's taking chunks of memory and moving it and rewriting it and this is the way that it tests it will do it in random patterns it will do it in actual patterns it will do it in random sizes small size big size so it's just taking chunks of information it's moving it all over the place until all of the bytes have been tested on the ramp all right so 26 minutes later we are on past two um no error errors were found so as it accumulates and does all these different tests the hammer test the sleep test it'll go to sleep wake it back up with a big load and then go back to sleep whatever sounds like i'm describing something but i digress it's gonna do it four times now we don't have a hundred minutes right now to spend on this but if you want to be completely thorough with your test you would just walk away and let it go and do its thing and if any errors are found well that's bad news more often than not you're not going to find any sort of errors here so i'm going to actually just go ahead and exit this test and you can configure this to let it go just one pass one pass is usually good enough but four passes would be the ultimate test now the more memory you have the longer it's gonna take so now that we've done all this this pc is now ready to be given to joe for his birthday and i have confidence that it won't have any sort of weird issues that haven't already been you know discovered and fixed so with that guys thanks for watching i hope this video has helped you feel a little bit more confident with building your first computer and making sure it all works properly share this video with someone maybe that's talking about their computer acting weird or weird stability uh issues and usually by doing these tests they can figure out what the issue is thanks for watching guys and of course we'll see you tomorrow [Music] back in five years [Music]

Channel: JayzTwoCents

Views: 1,069,694

Keywords: pc build, how to test a new pc build, how to stress test a pc, how to find bad pc parts, how to find a bad cpu, how to find a bad gpu, how to test a cpu, how to test a gpu, how to test memory, how to test pc memory, how to test ryzen, how to test intel, how to test

Id: AkNEvIWp2fM

Length: 19min 19sec (1159 seconds)

Published: Thu Dec 16 2021