My vote is bad power supply/power cables. Some games wont draw as much power so wont crash -maybe the early gen EVGA 3070 you tested doesn't draw as much power.
Also makes sense when jay puts the 3070ti on his test bench why it works…probably has a 1000-1300w psu on it which can handle that load being asked of it. Evga would put it on a test bench with I guess a 750 at least as well. I wouldn’t think they would match hardware to what you have…no budget for that.
Same thought, Ive had power supplies going bad create all sort of errors that pointed to different things, and sometimes particular hardware is more sensitive to it
Just had something like this on my 3080. I could not figured what was happened. Swap MOBO, same problems. At last and $350 decided to test all the cable from the PSU and one single cable was sending our 11.8 volts instead of 12 volts. Swap the PSU and problems resolved. Now, I have another z490 for a second computer.
I've seen similar issues due to bad Voltage drops in the GPU cables. In my case, the cables themselves were to blame. Bad soldier joints on the PSU end. Might be worth trying a different PSU.
What you can try is to swap the power supply together with the modular cables. Your own RTX3070 have a lower power limit than his FTW3. Another thing you can try is to limit the PCI-E to 3.0 instead of 4.0 to troubleshoot. Or up the CPU VSOC to 1.1~1.2V .
This. The first thing I thought of after Jay said the viewer had already RMA'd the motherboard and GPU, and had "swapped the CPU and RAM," was that the one thing that hadn't been replaced as part of troubleshooting is the PSU. It's as if the power draw spikes, then the GPU basically has a brown out and "disappears" from the system as far as DirectX is concerned.
I thought about this. The issue is if this was a power supply issue or transient spike it would take the entire OS down. Not throw the error you are seeing.
GPU`s of the same type do have different power requirements; and some PSU`s do have a not so Smooth power delivery which some sensitive cards highly clocked do not like.
@@KellicTiger not necessarily most PSUs have a dedicated 12v system just for the pcie connectors so if that Browns out it won't take out the 5v power to the mother board
@@KellicTiger Agree, I've had a power a PSU issue before and for me it most of the time just cut to black somethings when i was playing some games (namely Distance). All i had to do to test it was get OCCT (i think this was the software) and do a GPU test. Crashed like a charm.
Hi there, about 28 experience with IT support. I encountered something very much like this only a few times.Each time, it was with different hardware, but I first found a bug like this back in the Windows 98 SE days, and I have seen this effect up to Windows7. There is a rare situation where Windows gets confused in regards to hardware serial numbers and hardware product id numbers. In very rare situations, two bits of hardware can confuse windows because of the interplay between the hex values of the two bits of hardware. Normally this was between the motherboard and something going into the the expansion cards. You could put the expansion card into another PC, and there would be zero issues. You could put an identical card into the motherboard and have zero problems. But putting the two together created problems. The only workround round that I found to try and get the two to talk again was to alter the serial number information on the expansion card that was being put into the motherboard to get round this bug. I have had issues like this with gpus, soundcards and network adapters. It is very very rare. All of the cases I worked with, all would BSoD. But with Windows 10/11 it has better protection againest BSoDs. So instead of that your getting a crash. Its kinda like, a specific value is generated as a IO call, but due to this bug, instead of going to the correct place, instead because of the hex value sum, it infact gets sent to a different I/O address. I think the way windows is handling it is akin to it sends the data to the wrong place, and gets back a different result. So assumes that device has been unpluged. Hence error. I hope this gives you some food for through. Tho given bus sizes these days and the through put of data per-second. It would be very hard to catch the error in live data. Good luck. So far you have tried everything I tried in these situations.
I also strongly suggested trying with a different power supply, there are factors here that make a hardware hacker like myself suspect this could be a case of unintentional fault injection. By that I mean that the newer 3070 variant may be drawing enough power that the PSU is failing to be able to maintain a consistent and stable voltage to the power rails of the card. This can cause bit corruption and even screw up the instruction pointer in processors, either way the result is usually to expose undefined behaviour like this, most of which as in this case just straight up break what it was doing and nothing useful comes of it. Those of us that use it for hardware hacking live for the times when the undefined behaviour turns out to do an interesting and expose opportunities to encourage useful functions the designers did not intend, like when a well timed glitch causes the hardware to skip a check that was intended to stop one from customising firmware or whatever.
@@seraphina985 I went with my cut, and after I posted what I did, and I noticed other people was suggesting PSU. I was thinking to myself that is another good option. I think the main reason I went with my gut on this and posted what I did, is that in the video, I didn't hear any coil whine. Which normally a solid sign that the PSU is struggling. Even tho there is not coil whine, it might be a case that it cant deliver clean power.
You can watch Greg Salazar. He has a series where he fixes peoples PCs and it’s pretty much what you’d want. Different systems with different issues and tons of troubleshooting.
(PC-busters) If there's something strange In your computer Who you gonna call? (PC-busters) If there's something weird And it don't look good Who you gonna call? (PC-busters) I ain't afraid of no PC I ain't afraid of no PC If you're seeing things Running through your display Who can you call? (PC-busters) An invisible window Sleeping on your desktop Ow, who you gonna call? (PC-busters)
Please! Iv been battling a persistent file less hack n Iv already lost all my files cleaned my drives Iv reinstalled os 4 times internet no internet off usb. Seems to run through powershell…
Very often such issues are caused by power supply. Specifically as the user said it might crash only about 30 minutes into a game. So try another one. Alternative suggestion: Your test rig is running the card vertically while in the PC it's running horizontally. We have seen bad vapor chambers recently too. As well as even mechanical stress might cause the issues. Ah, and I suggest using memtest86+ and let it run over night to validate the memory - not only the memory modules themselves but also compatibility in an end-to-end test. I've seen many EXPO timings an recent Zen 4 systems booting fine into Windows but crashing on high memory load. Memtest86+ revealed sporadic memory errors. There should be absolutely NO errors reported on a 24h run. Even though this system might not suffer from memory issues as you already swapped the sticks and running JEDEC timings which should be pretty safe. Though still suggesting to test rather than try-and-error approach.
From the research I did, the 3070 FTW card has greater power requirements than a standard 3070. You most likely have a better/more powerful power supply in your test bench. I would try a bigger PSU in his machine and see if that solves the issue.
Five minutes into the video I was just saying "POWER. SUPPLY." every time Jay was trying something different. It's either PSU or (as he even mentioned!) a bad NVMe drive.
@RogerWilco99 are those 80+ gold? And like are they new or old? I ran a 4090 oc on a nzxt c850 with 0 issue or it even getting hot. I ran a r9 290x oc drawing about 300-350W with a older i7oc drawing like mid to high 200s on a gamedias 650W 80+ gold PSU without issues. If they're new those evga 80+ gold PSUs sound very Sus to me. And that one case with the gamedias PSU doesn't have good airflow either.
yeah I mean if the card works on the test bench fine and it's only that system I would guess psu not pumping enough juice for the card to keep running stable underload
Next time i would suggest to test the card on your testbench with the same orientation as in the case. Bad BGA contacts or broken traces/vias are more prone to show when the card is installed horizontally because of PCB flexing.
Thought this also, damaged/cracked PCB at the PCI-E connector or damaged/cracked BGA joint(s) near the connector. Vertical mount does not put lateral pressure in this area.
Hi. The card is vertical on your test bench and works fine. Do the same in the PC by tilting the case onto its side and test. Sometimes it's a case of a mounting pressure issue somewhere on the actual card.
@@RamboRune He says he came back to it and found it locked up, but it didn't have the same error, so who knows what happened while he was away, Windows updates etc, he mentioned a game install too. Could be the same issue without the error notification, or just some driver conflict while he was away. I believe the card ran fine vertically otherwise.
@@b.buster. A card that doesn't have issues. They're suggesting it's the orientation on the bad card, that maybe it's some weird pressure that's doing something.
Kinda cool to see someone as knowledgeable as Jay to come across an issue he's stumped with. Sucks for the owner though, I can't imagine how frustrated he must of been this whole time.
That happens all the time, I am playing with computers since like 2004 and still, I see problems which I have no idea what to do with that and those owners think I am magician and I have to fix it in 2 minutes, but it doesn't work like that, sometimes you just have to randomly replace all components because you just don't know where is the problem and sometimes problem will disapear and you have know idea how you fixed it and what was the problem, so you don't know if it's fixed permamently or not. 😀
i swear to god when Jay said "imagine we seeing this card go fooosh" the fire animation looks so smooth and clean, who ever is the editor kudos to you.
You should do more of these videos! Everyone has some type of issues and it was great to see you systematically go through all the problems to find out what the cause of the errors are! Keep up the great work!
But he didn't go through the necessary trouble shooting steps especially when knowing that the person have been swapping out components. He didn't rule out the motherboard wiring, the CPU, or the PSU after he noticed that the card worked on his personal test bench. And he can't compare a weaker 3070 card to the 3070 FTW3 when that card draws a little more power. I would've like to have seen the PSU.
@@spankbuda5760 I agree 100%. He didn't work through this problem at all with a proper scientific or troubleshooting method. *In the video* he states that he has to prove the defect both ways - good gpu fixes it in customer system, but also suspect gpu fails on known good system. He invalidates his theory that the GPU is the problem because it works well in a known good configuration, and then refuses to eliminate further variables or continue with scientific method. Jay found himself in EXACTLY the same place as the customer, because he didn't follow proper troubleshooting steps.
Weird (inconsistent) issues like that typically indicate to a faulty PSU. Another idea Ive just had: different Mainboards use different CPU-PCIe Lanes for the first PCIe slot, so maybe thats why the GPU works fine in your test-rig. --> reseat the CPU or swap the CPU?
your test bench is horizontal. If there's a bad solder joint on the card it might be orientation specific. So maybe try run his system with the system horizontal instead of vertical. Also his RAM was in the wrong memory slots which would prevent DOCP from working. EDIT: Like some other comments have mentioned it might be a PSU issue but I would've expected the GPU undeclock to fix that.
Yep, i was thinking the same as you. That underclock should have reduce spikes on the PSU. This remember me the strange issue i have with my setup. My display doesn't turn on after windows initialize the driver. I have to unplug and plug it back in. (Or disable the screen in device manager and re enable it) My brother has the same display. His work fine. In both mine and his setup. No idea what the issue is... And yes we swapped the GPUs too but my monitor work in his setup with my GPU. I have this issue since 2015/2016... I discovered it when i update my GPU from an hd6870 to a gtx1060. I have reinstalled windows multiple times, changed the drives, updated the drivers etc... It seems a specific issues with that monitor with that GPU (my brother has the 1050ti) and my MB/CPU/ram setup. Nothing to do.... Well except exchanging the monitors or running a simple script on autorun. (Yes my brother at the end didn't want to swap monitors)
yep Buildzoid you're absolutely right and i noticed and had the same toughts about ram, gpu orientation, and potential psu spikes or power drops issues . also ram in incorrect slots can sometimes create lots of weird issues in some cases .
my guess is, something on the motherboard is pushing against the GPU plastic housing and isnt creating proper contact with the PCIE pins in the slot. I say this because I have a Palit 3080 that due to its design, the frame hits my PCH plastic housing and prevents a fully seated GPU without putting some seriousssssss force on it just to get it too clip in, and even then it still somewhat pushes away from the PCIE slot so i know connection isnt 100%, wondering if something similar is happening here. DXGI errors are straight up device is no longer visible in device manager (he should check event viewer to see what occurs before that event happens) DXGI errors also have nothing to do with power (which should be ruled out via down clocking anyways) I think to rule out PCIE slot, he should just run it off a riser/extension cable and try both PCIE slots (looks like it has 2)
Yep, I had a VERY similar issue with my MSI Gaming X 3080, worked laying down , but in normal standing position, properly set and secured, would randomly crash. Got one of those GPU support things you can extend to support from the bottom of the case. Haven't had a problem since. I also do move my setup around a lot and drive it all over the place so maybe that played a part, since it didn't always do this.
As many other commenters have suggested, it may be a faulty PSU. I'd be interested in seeing that PSU tested. Possibly a ripple that the card doesn't like and causes it to be unstable, or the way the card draws power could be in such a way that the PSU can't sustain whatever demand THAT specific card needs? This is all speculation on my end, but I hope that you and the viewer can get their system up and running!
Actually he didn't do the necessary steps in the process of elimination like checking the motherboard bios, the motherboard wiring was properly connected, using a different PSU, or checking the CPU, socket, pins, and the cooler mount and paste.
In my case (DXGI on 2080Ti), it was bad memory stick. In beginning trying memtest end without errors, but after month of black screens and game crashes with DXGI and TDR errors, finding solutions and triing differnet fixies, i try memtest again, and voila. Errors in same memory address spaces. Change memory stick and problems gone away.
Check 3.3v resistance on the GPU. Low resistance in the GPU will cause lots of issues to the motherboard. It can corrupt bios, crash, corrupt ssd etc. Anything that uses 3.3v rail on the motherboard can be hurt by the GPU lowering 3.3v rail down to 300ohms or less. Other then temps, I'd look at that next. Best way to diagnose something like that is good old furmark
Out of what I call the big 3 for tech channels, JTC is the only one that does troubleshooting consistently. Having videos that show you troubleshooting a PC helps those who are new at this. Great content as always!
Two things, first I would try a different PSU. There could be a transient spike causing the GPU to crash under load, Gamers Nexus has talked about it a lot lately. The second being trying to update the motherboard BIOS.
Bad transient spikes will completely shut down the system as shown by GN, because it triggers the OCP protection of the PSU. Here the system remains on and functional. The issue could be PSU or cable related, but it doesn't seem to be caused by massive transients.
@@thelegendaryklobb2879 however it could cause the card VRMs or mosfets to recieve 'dirty power' and if they are strong enough or the PSU isn't strong enough, could cause weird hardware software issues on other hardware. Power is almost ALWAYS my issue on my customer's systems. Whether a bad supply, bad connection, or even just bad power from the wall
i had the same issue... i replaced psu, ram, drivers, changed clock settings, windows etc etc i am as stumped as him. I could relate to his frustration in this video
I thought of the Mobo BIOS too, but it sounds like the user also tried replacing the board. Still wouldn't hurt to check, which I have a feeling Jay did.
I agree with other comments. I have run into similar issues that turned out to be a power problem. I would say the power supply or cables are failing. If you do find the solution please keep us updated.
As other comments I've seen have pointed out, I'd suggest trying a different PSU. The older lesser 3070 would be drawing less power than the FTW version. It would also explain why only some games crash since different games utilize cards differently and can cause much different power draw situations.
on that end surprised no cinabench test. also why no driver booster check? there could be old drivers else where on the system causing a long enough hang for it to only fail on a few games/situations. there's one ive been avoiding: Amd PSP updated 9/16/22 because for whatever reason it causes only my discord and vlc media to error out and crash once its installed. might be something similar.
I'd agree, he could monitor voltages to the card in HWInfo, I'd think if the PSU is not up to snuff you'd see voltage drops.. I had an issue where my machine was crashing, turned out the extension cables I used to dress up my PC were dropping the voltage down to about 11 volts under load on both the 24 pin and PCIe cables. It drove me nuts figuring it out til I saw the drop and the light bulb went off.
I would def try this as I had a bad power supply many years ago. Everyone kept saying that a power supply would not crash the games or the system, but it did. When I played a game it would crash, the hdd even started knocking. Changed power supply - no knocking, all games ran fine.
Most everyone's saying it's likely the PSU and I'd agree with that theory. Another approach I would take is matching CPU (or as close as possible) on the test bench since it could be the interaction between the two that causes the problem.
@Vercusgames i wouldn't have thought PSU either, but another comment suggested that maybe the card jay tested was older and thus not needing as much power. I don't think they mentioned his PSU in the specs which leads me to believe Jay didn't think it was under supplied.
Hey I had a similar issue with one of my 3070s. The card would work in a vertical mount versus the standard PCIe socket. Note with your experience it worked on your test bench when it was in the vertical position. Assuming you've tested the PSU.
Jay, I had 3 devices now that I have helped people with with almost this exact error. And the error kept getting worse until the system wouldn't even boot (months). All the systems had 30 series cards and rm series Corsair psus. On all of them, swapping the cables to different psu ports or removing extensions fixed the issue. Probably why it works in the other system. Please try to swap the psu
I have a 3080 ftw3 from evga, R9 5900xand a 850rmx and I too had very bad crashes , until I turned off DOCP, and unplugged/plugged in the PSU… and now my games run well
This is amazing to see! I've has a Possessed PC from Hell for 5 years and to see someone like Jay take a random PC and step by step go through his 2 cents on probable causes and then test it! This is REAL content here. I LOVE IT! Please do more problem solving like this (Jay and Other Creators). Thank you again.
I'm surprised he never tested the PSU as it is definitely one of the possibilities. The GPU probably has a higher power draw than the replacement card, even if they are reportedly the same on the box, which is why it worked. The GPU starves and crashes during transient spikes. Could be the PSU or cables but at this point almost certainly a power delivery issue. Underclocking was certainly a good test but doesn't eliminate all power related issues.
Power issues cause all sorts of issues in every field of electronics. Noise on radios, corrupted SD cards, random communications failures, Raspberry Pi or computer reboots. It's especially liable to get glossed over because it's not flashy. Until it explodes, that is.
Phil you almost gave me a heart attack with that flame on the card and it's not even my system. That was a beautiful edit. This was an excellent Troubleshooting 101 type video using a real world scenario. I know this isn't what you want your channel to be, at least to this extent, but I think this is one of your most valuable videos yet for the average PC user to watch.
This is a pretty interesting concept for a video series. On one hand, having a number of different videos that more or less match the search terms that people might look for when they're having computer problems means that you could effectively create a series where you do more or less the same thing and follow the same sensible troubleshooting process. That would make people more effective at troubleshooting their own machines because the biggest problem with folks is that they just try to skip through. Doing the same thing over again and finding different problems in different viewers' machines may help reinforce actually useful troubleshooting methods. On the other hand, people are idiots and people with tech problems that your videos might help may expect you to help them more in the future. Tricky thing to balance.
theres another creator who does something similar called fix or flop where he does exactly that but the titles arent really in line with an easy troubleshooting catalog
My initial guess, just off given info so far, power issues. Transient spilke from the GPU, but not quite enough to crash the system, or perhaps even a short in the cable to the card. Maybe even a small short on the board itself. Strange multiple cards would do it though.
Jay if you were looking for some fresh type of content that can do well, I think this video is a good start. Interested in more of these, entertaining and knowledge gain at the same time 🤘
I would probably try changing PSU... unstable power (especially on higher draw) could lead to all manner of unstable behaviours. And that is one component that is often overlooked when looking at the stability. And it can be possible that this specific piece of card is more prone to crash on power fluctuations, while others are less and just work.
I had a older PSU that kept crashing my whole system for a while so eventually i suspected it may have been PSU so i bought new one (modular this time) and tested it and beside some random startups after shutdown for few days (then it somehow stopped booting up again after shutdown) it works fine so far. This issue definetly seems like PSU or GPU power cable.
I've had parts that just didn't want to cooperate together but worked perfectly fine when separated. The more complex our technology the more variables to consider.
I once saw someone who had a PS3 console that would work with any tv but theirs, and a TV that would work with any PS3 but theirs. No one at Samsung customer service could figure out why that happened
@jay The DXGI error just means the application lost access to it's rendering target, when it says "device" it's not talking about the actual GPU but the virtual device that the DX pipeline targets for rendering. DX / OpenGL rendering does so to a virtual target that the drivers then push to the real screen, it's how we you can alt-tab out of fullscreen and windowed rendering works. That would only happen if something bad happens to the graphics card itself, my suspicion is the PSU's power delivery isn't clean, different cards have different tolerances and chances are that specific model card is very intolerant of out-of-spec power.
I would try a different PSU. His 3070 might be pulling a little more watts due to the higher clocks. Probably won't do anything but worth a shot. Seems to be the only common denominator in his build
That's where I was going too. In my experience weird, random, inexplicable nonsense like this is scarily often a power issue. I always seriously over spec my PSUs in my builds. Power scares me 🙂
My Sapphire Nitro + 6800XT was having DXGI issues as well. Swapped in a Corsair RM850x and bam problem solved. I used HWInfo to log everything while it would crash and noticed literal millisecond voltage drop "gpu disappeared" then the game would crash. Apex was the easiest to get to recreate the problem. I could play Cyberpunk for an hour before it would happen. Drove me crazy!!!!!
@@singlsrvngfrnd yea I have a be quiet dark power 850w psu. Seems to work well with my 6900xt. But I've had thar psu for the past 3 builds. Not sure how many years it has left.
It's almost certainly the PSU, but given Jay under-clocked the GPU and it still had issues I think it might specifically be transients that are causing the issues. IIRC Apex has been said to create transient spikes more frequently, and the 30 series GPUs were notorious for high transients too, so if the PSU dropping is voltage in a transient spike that would certainly explain the behavior shown.
Honestly these types of troubleshooting videos are the best thing ever made. If it wasn't for TH-cam Troubleshooting videos I'd have "bricked" computers because I just wouldn't know what to do.
I would say to try different power channels if he has a modular power supply or do a vertical mount because in your test rig, it worked fine when it was vertically mounted. It might have a problem when it's horizontally mounted.
My daughter had a very similar situation with a Gigabyte 3070. It would routinely crash when she tried gaming on it, just like the one in the video. We put it in my system to test like Jay did with putting the one in the video in his test bench. It worked fine in my rig for a few days and then started displaying the same behavior. I had tried the Bios and firmware updates just like Jay did. We ended up returning that card and she got an Asus card which has ran flawlessly.
PSU and(or PSU cable problems are a fair suggestion though undervolting should cover some of those. I'd try in this order: 1. Severely underclock the GPU and VRAM under 1 GHz, maybe some undervolt too 2. Check/clean PCI E lanes on the card and pins on the mobo 3. Downgrade mobo's BIOS (for the sake of it) 4. Get a priest
that's what i was thinking, throw msi afterburner or asus gpu tweak III on the machine and start with lowering the power target and maybe the clock just a smidge and see if it persists
i would love to see more videos like this! my pc has a new problem every couple of months, so seeing how you fix problems would be awesome. im also everyone's tech support
PSU, potentially the launch card had a lower TPD than the later (factory boosted) FTW3 card, so when put in the viewers system, the launch card could perform ok. Then when the FTW3 GPU went into Jayz bench, which presumably has a pretty robust PSU, it was able to fully boost.
@@zalatos possibly he adjusted the clocks, not the power draw. While it might have naturally reduced, depending on how the BIOS is written, it might still ask for the same amount of power when it boosts.
Like several others have suggested, it could be a power supply or cable issue. Trying a different set of power cables for the GPU would be one place to start. Even though the other card worked, it's probably pulling less power then FTW card. Also as others have said, switching the orientation on the test bench to match how it's installed in the PC and see what happens...
@@DarkReturns1 I am a little confused by your comment. There are several companies that make custom replacement cables for a lot of the different brands of power supplies. Now I could understand if you tried to use a cable from a different brand entirely as that may have different pinouts, etc. Even different models from within the same brand. But using a replacement from the same brand and model or using one custom made for your exact power supply shouldn't be an issue.
Weird issues like this can sometimes be caused by a sub-par PSU, might be a transient spike or something that the PSU can't handle properly causing the GPU to not have the voltage it needs and not operating correctly (resulting in a graphics API crash). It would also explain why the GPU works in the bench (better PSU that can handle the spikes) and why another GPU (that might not have as high spikes) could work in the PC. It might also be worth looking at the event viewer logs to see if there is anything sus going on in there, but given the clean OS install I'd still put my money on the PSU.
Lots of people commenting the PSU, but if you google “apex legends DXGI error device hung” you’ll find endless posts over years of time about this specific crash in Apex. It effects thousands of systems, every config under the sun. A bad PSU wouldn’t explain all of that, there’s no way that many people have faulty PSUs. Also worth mentioning that I’ve dug through dozens of these posts and videos because I had this problem myself with Apex, never once have I found a real solution including PSU/cable swaps. Also doesn’t make sense that it effects Apex specifically, so prolifically. Meanwhile myself and basically everyone else who experienced the problem have never had the issue with another game. Obviously I never found a fix, so I just stopped playing the game for a month or two. Then I went back to it and tried it again and it just worked fine… really weird.
@@charlesbrown4483 DXGI errors can happen for many reasons I'm only suggesting that it is the PSU in this SPECIFIC case because of the behavior of it based on the troubleshooting done. It is possible in your case it is caused by something else. It makes perfect sense that it affects one game (APEX) more than others. Each game is different, each will put different loads on different hardware component at different frequencies, so if there is a specific issue that happens only under specific conditions then it would only happen if a specific game creates those conditions (which seems to be the case with APEX). A similar thing happened with 3090s which had issues in New World but were perfectly fine in most other games. It is quite possible that APEX is creating GPU power spikes and the PSU isn't properly handling them which could result in voltage drops causing instability. There are certainly other potential issues but based on the GPU working fine on the bench system, a different GPU working fine in the PC, and the issue persisting between mother board replacements and clean OS installs, I think the PSU is the most likely cause in this particular case.
@@charlesbrown4483 It is also worth mentioning that it isn't that PSUs are faulty, it would likely be that certain PSUs just can't handle the transient spikes created by the GPU to handle the load APEX puts on it. There is a good Gamers Nexus video about transient spikes (which were quite problematic in the 30 series) and GPUs were pulling 2-3 times the normal load. So if the GPU normally pulls 250w of power then a transient spike could pull 750w briefly and not all PSUs would handle that, especially if the total wattage of the PSU was less than that, a 750w PSU would normally be a reasonable PSU for a 250w GPU so it wouldn't surprise me if a large number of PSUs weren't able to handle the spikes properly, especially cheaper PSUs with smaller/cheaper capacitors.
@@AusSkiller Okay yeah I’m with ya, that’s a good theory. Though if we’re talking about transient spikes caused specifically by playing Apex, as you mentioned that’s not a faulty GPU. Which is the conclusion most people are coming to. So imagine someone has a 750w PSU, they take the advice of the comment section assuming their PSU/cables are bad and just replace the unit with either a new in-box model of the same unit, or just some other 750w unit. In theory there’s no reason that would fix the problem. They’d have just wasted money on a new PSU and should still be getting the crash. So if this is the case, then the comment section’s advice should be “upgrade to a much higher wattage PSU” instead of “your PSU needs to be replaced.” Though I gotta be honest I have little faith in any solution for the problem. It’s like the zodiac killer’s puzzle of PC gaming lol.
@@charlesbrown4483 It entirely depends on the 750w PSU, some will be fine with transient spikes up to 1500w, others might struggle with as little as 1000w, and the length of the spike or frequency that they occur can also be factors for if a PSU will be ok. That's part of the problem with the transient spikes most PSUs aren't really rated for them so it is difficult to know which can/can't handle them, and also what kind of transients a component can produce, so it is near impossible to know what you actually need. My advice for anyone with an issue like this who thinks it might be the PSU is to find a PC repair place (or better yet a PC enthusiast friend) and see if they can test the system with one of their PSUs that is either higher quality or higher wattage to see if that fixes things. If it does then they'll know the PSU is definitely the issue and they can buy that kind of PSU to fix it, if not they might have a different underlying cause.
Agree with others that have already said this. Check power cables/power supply. Since it worked in your test bench, I would say the issue is potentially one where the card is experiencing either a power event or thermal event (test bench is open air so could allow better heat dissipation.
It would be good to use the performance overlay while in the game so you can see the info on the components before it crashes. So you can see if it some kind of spike compared to your other card that does not crash.
Had almost the same issue as well on an older FX8350 build where it was crashing while loading games, when just the day prior it was running fine. I noticed that the power supply voltages were fluctuating quite a bit. Swapped out the PSU and it has been working perfectly since.
Would love to see more videos like this, modern builds with puzzling issues and how you got around to solving it! Now im eager to see the next video from this and what the actual issue was and how you were able to help solve it for this channel follower! (I'm leaning to the commenters side on this and saying psu/psu cables)
I had the same thing happening with my system. It ended up being the NZXT C650 Power supply went bad. The PCIE portion of the PSU failed. Swapped to a Corsair RM750e and no problems since.
The reason why the GPU works in the test bench is that there the card is not horizontally mounted, which leads to the bad solder ball joint/joints in the affected GDDR6 memory chip to have better contact. The root cause is GPU sag. This is quite typical with these bigger and heavier cards when you have them unsupported for a long time --> the PCB bends and the VRAM chips closest to the PCI-E slot ecentually get bad solder joints due to the mechanical forces. A MATS/MODS test with the card horizontally mounted would reveal the chip with the bad solder joints
holy shit, so ive been having this device removed thing happen, and when I read your comment it hit me, on my case i have a thing to hold up the gpu so u dont get this, but when i upgraded to a 3080ti it was so big i had to remove the thing holding the gpu, and i couldnt figure out how to screw the gpu sag device farther back so i just didnt use it, do u know of any other ways to solve this? maybe ill get some ties or something to hold it up, hopefully Jay reads your comment cause this is likely whats causing it.
as someone who troubleshoots & repairs machines for a living, i really appreciate seeing another professional going through the analytic cadence, and even into the 'this should not have any effect on the fault symptoms, but i got nothing else left to swap out' area :D i'd live to see more of the cursed pc troubleshooting
What crazy to me is he completely missed troubleshooting steps that we immediately obvious to me and others in the comments. This would be faulty PSU cables or PSU itself.
i mean, it's also a triage in there. when you consider the symptoms, and get a shortlist of possible causes, most people will ask 'which of these is simplest and fastest to test?' and then work their way up from there. but what is considered simplest by one person is not necessarily same to the other
@@Kholaslittlespot1 I literally wasted months or even years by trying to solve some PC problems, I think it's more like addiction, similar as smoking or playing slot machines, I really have to stop with that and throw all computers to dust bin, becaue I am wasting my best years of life because of that. 😀
@@Pidalin haha I feel you, but I like to think of it more as a passion! In fact, it's helped steer me away from several dangerous addictions in the past. Of course it still needs to stay in balance. I used to have a whole room full of old machines that I eventually gave away to charity... Then again, I still have a habit of picking up old commodore Amigas and ZX spectrums from eBay with the intention of fixing them up, so I'm still pretty bad! xD
The only two things I could think to swap out at this point would be the power supply, because maybe there is something going on with the power rail for the GPU. The card you swapped in wasn't an exact copy of what was already there, so that could explain why yours works in his system, and his works in your system. The other thing I could think of would be try swapping the hard drive itself. You mentioned that there could be an issue with the controller overheating. While I don't believe that to be the case as the system is crashing rather quickly, you may still be onto something with thinking there is an issue with the hard drive itself. Out of the 2 things listed though, my money would be on the power supply.
I remember about a year or so ago that you had a video showing some weird issues with the motherboard bios not liking the video card which caused weird crashing and other issues. Even if this one is on the latest bios, I wonder if a reflash would do anything. After that, my auto mechanic experience kicks in and I'd start looking at physical things, cables and such
2:23 as an IT professional with several certs & almost 20 years under my belt, I appreciate this part. I can't tell you how many people have given me the shocked 😳look when I mention event viewer or looking at logs.
I had Same issue with several games and i had a burned Mosfet for some reason gave it to be fixed and it works now (mine was already a long time after RMA time)
Like other people said, i would try with a differnet psu but before that i would look inside the bios and make sure resizable bar is turned off. Maybe some weird bug in this particular combo of MB / GPU does not like it.
Had similar issue with my Sapphire 5700xt. Worked fine with other system except the new build. My case was cause of bios. After tested with several old bios version of motherboard which is asrock, I was able to fix the problem.
From my experience the DXGI issues are from OC on the videocard, happens mostly on EA games Later Edit: as seeing the card been tested on a different system and working I can only think about the power supply. There was a discussion regarding the peaks of 30 series wich could imply crashes
Ive had this issue. I worked jt down to my oc in msi afterburner. At times it runs fine and then and would randomly crash during an intense fight, sometimes in firing range it would do the same with movement shooting dumies. This is all in apex using msi 3070ti ventus @ 165/500+ clock. Took it back to stock snd just adjusted fan curve
I'm leaning toward a bad psu or psu connector. If it works in your test bed, then it might be whatever is supplying the GPU. Since virtually every other part has been replaced, then the only two places to look are the GPU and psu. The gpu works in your testbench (which has a beefier PSU). The only option in my mind is his PSU.
I think putting one known working ram stick in every ram slot on the motherboard will 100% rule out the ram problem (although the memory channel of the cpu can make things more complex). The second thing I thought of was just clearing the cmos. It is very easy and fast to do, even though that might not be the issue. Same goes for testing ram I mentioned before, fairly easy to do and won't take a lot of time.
One of my old desktop had the same issue but by doing a base clock OC it sort of fix it for a few months before The PUS go kaboom and bring the rest of the parts with it.
Maybe it's worth checking the PSU. The PSU might not be a high enough wattage or starting to fail. The GPU during these games might be trying to pull too much power for the PSU and when the GPU doesn't get it, it becomes unstable or crashes temporarily to cause these said issues.This would also explain why it works in your system and not there's. Not entirely sure that's the issue but it'd be the next thing I would consider looking at.
this was really fun to watch, id love to see some type of mini series on this stuff. also my vote (as someone else said) is on it being a cord/power supply issue, its the only variable that never changed between systems that is literally directly connected with the gpu. both cards you swapped out i believe have lower power draw so maybe the reason they didn't crash and the good cables/psu in your other system ran the card fine.
Jay, move the M.2 to the location ABOVE the video card. Also, the last 3 factors that are remaining as common are the case, SSD, and the POWER SUPPLY! If the supply is possibly defective or weak on a power rail, it is essentially overloading for a second and tripping a reset internally (I am chucking spitcannonballs here trying to help). If his supply is borderline for power, or a cheaper brand, it may not be able to handle the load.
So - the one item that caught my "eye" was the PSU - Corsair RM850 - I had one of these a few years back and had all the same type of random issues - in the end, replaced everything in the system except for the PSU and still had the same issue, replaced the PSU with the same from Corsair RMA and that one worked for about 6 months and then did the same thing, so got rid of it and went with a EVGA 750 and boom, it was fine - worked like a champ from then on. In fact that system is still running great for my daughter years later.
Dang - I returned a Corsair refurbed RM850 to the store (good store, no problems) about two months ago. Was progressively failing earlier and earlier, until system usually could not boot successfully.
I've ran the RM850 for years without issue (It's actually been in 3 different builds so far). you probably "lucked out" and got a bad one.... OR you had dirty power from the power company and it caused the issues. I have never had a Corsair PSU fail for me either, so hearing someone have that kind of problem leads me to believee that the problem was not the PSU, but the power coming into the house, or something else related to it's environment.
@@AKStovall That doesn't mean anything. Just because you never had a problem with a certain brand doesn't mean others haven't. Errors happen all the time in manufacturing. It's usually a small number and probably not reported on much but it does happen, like the melting 4090's, or the incorrectly installed capacitors on motherboards.
I just saw this vid, and I was smiling. yeah... sometimes strange things happen. some weeks ago I had to upgrade a comp of a friend of mine... no matter what I did, the comp did stop every time... finally I did smell a burning air ( shortcut smelling)... it was a harddrive with a broken power connector. A peace of plastic between the poles was gone.. It's one of those things you never thought such things could happen, but it kept me busy for at least 40 hours. I don't know if you already solved the problem, but I am sure you did..
I'd be curious to see what happens if you try hooking up another power supply to his system. Could be that his card being the FTW3 is tripping a bad power supply whereas your card is not pulling as much power since it's an xc3 3070.
I had a similar issue; my PC would crash only during gaming (randomly). I replaced everything but the CPU. I then one day ran Prime95 longer that did before (because I fell asleep and walked off) and it gave me an error on just one core about 18+- hours later. Had AMD replace my 5600x CPU that fixed it. I normally stress test all my hardware 24 hours minus the GPU/CPU for only 30 minutes and 2 hours of Haven on loop. I am sharing this hoping it helps others with isolation of problems. Please run Prime95 for 24 hours part of their burn in on new builds/overclocks.
It's a very interesting issue to run into. It's just that card that's causing issues in that particular system. I had seen another comment where someone recommended checking the power supply, I would try another GPU that has a higher power draw and see what happens. Also, checking eventviewer will show you what kind of issues there are and give you a tiny bit more insight on what's going on (very rare occasion that it helps). Any other steps would be very involved and time-consuming, but you never know what could happen until you try them out! Great job on the troubleshooting, I'd pay good money for your service any day!
As many here, I vote for the PSU because I had some strange issues recently with this part and it looks like this computer had the same. I don't know the specs of your backup GPU but it may ask a little bit less power or it is asking for it a more stable way. Anyway, this is why everything works fine when you switched it. My own experience : until mid-2021, I had a EVGA SuperNova 650-P2 PSU with a i5-4590, 32 GB DDR3, a motherboard and a Gigabyte 1070 TI. I wanted to upgrade it to a Ryzen 3700x on a Gigabyte B550 AORUS Pro and more RAM. The GPU would stay because shortages at the time. The EVGA 650-P2 was recent because the Corsair I had died one night, and we can safely say that this is high-quality stuff. I NEVER spare money on the PSU, better safe than sorry. But the new parts were never able to go to POST. I sent back first the CPU (because LEDs on the motherboard pointed at it) and got a replacement who didn't work better, so I replug back my old parts and sent the new motherboard back to buy an another brand... but the issue was still here (with a difference : I could start it, but not more than one or two minutes before the screen goes black). Strangely enough, I plugged back the old parts and they worked fine again. 2 motherboards, 2 CPUs... so it couldn't be anything else than the PSU at this point. I asked a friend to lend me his own unit (sorry for the disassembly moment my friend...) and tested my theory : it was indeed the PSU. The new parts weren't supposed to ask more power than before, but my EVGA PSU wasn't unable to provide it. RMAed, the replacement unit was fine. Event decent PSUs can be faulty, sometimes in strange ways.
Thanks for sharing this content, Jay. I always struggle with troubleshooting but you have given me more ideas to look at next time I troubleshoot a PC!
I would be trying that GPU with a different PSU with different cables, and comparing pins on the motherboard and GPU. Seems more likely to me that it'd be a perfect storm of tiny problems that combine into an issue that isn't replicable with any other hardware combination.
@@zalatos yes, even in spite of an RMA’ed motherboard. For one, it’s not guaranteed that the user was absolutely correct on the course of events, and for another, it’s just a look at the contacts. Seems wildly dense to not tames few minutes to take a closer look at them even if that’s not the problem.
I had the exact same problem. I did all the things that you did, and was never able to resolve it. I finally just ended up buying a new card, and all has been fine since.
Squad is typically a fairly stable game, too... But random crashes with no discernable cause was linked to my PSU getting tired, especially on games that could spike the framerate extremely high. The 550 watt PSU might not be able to handle the current spikes from the GPU. The FTW3 might be pulling just enough power over the base 3070 that it can't keep up. Try a second PSU running just the GPU to see if it'll stay running. I had to do that for my 3090 while I waited for a replacement PSU to come in, lol. Edit: misheard it. It's an 850w PSU, but still sounds like a power issue.
I think you hit it on the head. A simple look on the web shows that the 3070 ftw3 should use a minimum 650 watt PSU. Considering how much RGB the user has in their system, they're definitely having a power problem where the GPU is trying to suddenly draw huge amounts of power and is unable to and probably causing the power to cut off to the GPU, which in turn would make the system think the card has been removed.
Yes, I misheard the first time, it is an 850 according to the closed captions. New models making me hear things, lol... I still think it's a power issue, though. May be a bad solder job on the card, though.
You should do more of these ''Computer from Hell''. I'm riveted at the edge of my seat and dying of laughter at the same time. Super interesting! I want more
Some tips: 1) Some issues can be solved by using different video output ports on the card i.e. using another DP or HDMI output. (try them all because some ports share silicon, think of it as usb slots sharing same USB hubs) 2) Trying using different PCIe slot and one that has a 8x electrical slot instead of x16 and/or lowering to PCIe 3.0 (or down to 2.0) might help with signaling, if signaling is the issue that corrupts. (if this solution helps then a workaround might also be lowering the motherboard PCIe buss speed, sometimes 99MHz helps when normal is 100MHz, or go even lower) 3) Perhaps it is a physical problem, If there are bad soldering you can perhaps inspect the card and maybe bake the card. 4) Lastly a long shot, have a look if there are no extra ATX standoffs creating this issue.
Being in the market of pc repair for over 17 years, I have seen this issue with many Nvidia and AMD gpus. And I have always found swapping out power supplies to be the best and last guess in troubleshooting DX and DXGI errors. Sometimes Pci-e power cables just go bad. And it's frustrating.
My vote is bad power supply/power cables. Some games wont draw as much power so wont crash -maybe the early gen EVGA 3070 you tested doesn't draw as much power.
I was going to post this.
Also makes sense when jay puts the 3070ti on his test bench why it works…probably has a 1000-1300w psu on it which can handle that load being asked of it. Evga would put it on a test bench with I guess a 750 at least as well. I wouldn’t think they would match hardware to what you have…no budget for that.
yeah, this make sense
Same thought, Ive had power supplies going bad create all sort of errors that pointed to different things, and sometimes particular hardware is more sensitive to it
Just had something like this on my 3080. I could not figured what was happened. Swap MOBO, same problems. At last and $350 decided to test all the cable from the PSU and one single cable was sending our 11.8 volts instead of 12 volts. Swap the PSU and problems resolved. Now, I have another z490 for a second computer.
I'd watch the hell out of a series where Jay travels the world cleansing haunted PC's.
And replace the Holy Water with Holy Mineral Oil...lol
With Henry Cavill as Jay. 😁
Ghostbusters/ Phasmophobia pc edition 😂
With Tech Jesus in tow
Excorsizing the demons.
I've seen similar issues due to bad Voltage drops in the GPU cables. In my case, the cables themselves were to blame. Bad soldier joints on the PSU end. Might be worth trying a different PSU.
Bad solder joints in the PSU or the cable?
I was thinking psu too dropping voltage
Then why isn't it doing it on any GPU installed?
i had the same issue and was the psu himself due to power spike created by the gpu
Yeah. My brain went also with psu.
Any updates on this? I’d love to see if you’ve solved the issue.
Has there actually been an update on what the fix was? I like others it seems were shouting PSU aha
It's the psu.
What you can try is to swap the power supply together with the modular cables. Your own RTX3070 have a lower power limit than his FTW3. Another thing you can try is to limit the PCI-E to 3.0 instead of 4.0 to troubleshoot. Or up the CPU VSOC to 1.1~1.2V .
This. The first thing I thought of after Jay said the viewer had already RMA'd the motherboard and GPU, and had "swapped the CPU and RAM," was that the one thing that hadn't been replaced as part of troubleshooting is the PSU. It's as if the power draw spikes, then the GPU basically has a brown out and "disappears" from the system as far as DirectX is concerned.
I thought about this. The issue is if this was a power supply issue or transient spike it would take the entire OS down. Not throw the error you are seeing.
GPU`s of the same type do have different power requirements; and some PSU`s do have a not so Smooth power delivery which some sensitive cards highly clocked do not like.
@@KellicTiger not necessarily most PSUs have a dedicated 12v system just for the pcie connectors so if that Browns out it won't take out the 5v power to the mother board
@@KellicTiger Agree, I've had a power a PSU issue before and for me it most of the time just cut to black somethings when i was playing some games (namely Distance). All i had to do to test it was get OCCT (i think this was the software) and do a GPU test. Crashed like a charm.
Hi there, about 28 experience with IT support. I encountered something very much like this only a few times.Each time, it was with different hardware, but I first found a bug like this back in the Windows 98 SE days, and I have seen this effect up to Windows7. There is a rare situation where Windows gets confused in regards to hardware serial numbers and hardware product id numbers. In very rare situations, two bits of hardware can confuse windows because of the interplay between the hex values of the two bits of hardware. Normally this was between the motherboard and something going into the the expansion cards. You could put the expansion card into another PC, and there would be zero issues. You could put an identical card into the motherboard and have zero problems. But putting the two together created problems. The only workround round that I found to try and get the two to talk again was to alter the serial number information on the expansion card that was being put into the motherboard to get round this bug. I have had issues like this with gpus, soundcards and network adapters. It is very very rare. All of the cases I worked with, all would BSoD. But with Windows 10/11 it has better protection againest BSoDs. So instead of that your getting a crash. Its kinda like, a specific value is generated as a IO call, but due to this bug, instead of going to the correct place, instead because of the hex value sum, it infact gets sent to a different I/O address. I think the way windows is handling it is akin to it sends the data to the wrong place, and gets back a different result. So assumes that device has been unpluged. Hence error. I hope this gives you some food for through. Tho given bus sizes these days and the through put of data per-second. It would be very hard to catch the error in live data. Good luck. So far you have tried everything I tried in these situations.
we need to thumb this up so jay actually sees it
Oh, wow, yes, I remember managing to get one of these bugs. Yup, editing onboard numbers were what solved it.
Bumping this up!
I also strongly suggested trying with a different power supply, there are factors here that make a hardware hacker like myself suspect this could be a case of unintentional fault injection. By that I mean that the newer 3070 variant may be drawing enough power that the PSU is failing to be able to maintain a consistent and stable voltage to the power rails of the card. This can cause bit corruption and even screw up the instruction pointer in processors, either way the result is usually to expose undefined behaviour like this, most of which as in this case just straight up break what it was doing and nothing useful comes of it. Those of us that use it for hardware hacking live for the times when the undefined behaviour turns out to do an interesting and expose opportunities to encourage useful functions the designers did not intend, like when a well timed glitch causes the hardware to skip a check that was intended to stop one from customising firmware or whatever.
@@seraphina985 I went with my cut, and after I posted what I did, and I noticed other people was suggesting PSU. I was thinking to myself that is another good option. I think the main reason I went with my gut on this and posted what I did, is that in the video, I didn't hear any coil whine. Which normally a solid sign that the PSU is struggling. Even tho there is not coil whine, it might be a case that it cant deliver clean power.
I hope this becomes a series. Such a great way to see how to best troubleshoot different issues.
You can watch Greg Salazar. He has a series where he fixes peoples PCs and it’s pretty much what you’d want. Different systems with different issues and tons of troubleshooting.
(PC-busters)
If there's something strange
In your computer
Who you gonna call?
(PC-busters)
If there's something weird
And it don't look good
Who you gonna call?
(PC-busters)
I ain't afraid of no PC
I ain't afraid of no PC
If you're seeing things
Running through your display
Who can you call?
(PC-busters)
An invisible window
Sleeping on your desktop
Ow, who you gonna call?
(PC-busters)
Please! Iv been battling a persistent file less hack n Iv already lost all my files cleaned my drives Iv reinstalled os 4 times internet no internet off usb. Seems to run through powershell…
@@VarkorothI was going to suggest him also. Probably my favorite tech TH-camr.
totally
Very often such issues are caused by power supply. Specifically as the user said it might crash only about 30 minutes into a game. So try another one.
Alternative suggestion: Your test rig is running the card vertically while in the PC it's running horizontally. We have seen bad vapor chambers recently too. As well as even mechanical stress might cause the issues.
Ah, and I suggest using memtest86+ and let it run over night to validate the memory - not only the memory modules themselves but also compatibility in an end-to-end test. I've seen many EXPO timings an recent Zen 4 systems booting fine into Windows but crashing on high memory load. Memtest86+ revealed sporadic memory errors. There should be absolutely NO errors reported on a 24h run. Even though this system might not suffer from memory issues as you already swapped the sticks and running JEDEC timings which should be pretty safe. Though still suggesting to test rather than try-and-error approach.
From the research I did, the 3070 FTW card has greater power requirements than a standard 3070. You most likely have a better/more powerful power supply in your test bench. I would try a bigger PSU in his machine and see if that solves the issue.
this^
Five minutes into the video I was just saying "POWER. SUPPLY." every time Jay was trying something different. It's either PSU or (as he even mentioned!) a bad NVMe drive.
@RogerWilco99 are those 80+ gold? And like are they new or old? I ran a 4090 oc on a nzxt c850 with 0 issue or it even getting hot. I ran a r9 290x oc drawing about 300-350W with a older i7oc drawing like mid to high 200s on a gamedias 650W 80+ gold PSU without issues.
If they're new those evga 80+ gold PSUs sound very Sus to me.
And that one case with the gamedias PSU doesn't have good airflow either.
yeah I mean if the card works on the test bench fine and it's only that system I would guess psu not pumping enough juice for the card to keep running stable underload
My exact thoughts without knowing the power reqs after hearing the circumstances of this issue.
Next time i would suggest to test the card on your testbench with the same orientation as in the case. Bad BGA contacts or broken traces/vias are more prone to show when the card is installed horizontally because of PCB flexing.
This
Came here to say this. The difference in orientation is one easy to test difference between the test rig and the problem PC.
Thought this also, damaged/cracked PCB at the PCI-E connector or damaged/cracked BGA joint(s) near the connector. Vertical mount does not put lateral pressure in this area.
Hi. The card is vertical on your test bench and works fine. Do the same in the PC by tilting the case onto its side and test. Sometimes it's a case of a mounting pressure issue somewhere on the actual card.
11:44 its vertical and still crashed it looks like when installed on the test bench
@@RamboRune He says he came back to it and found it locked up, but it didn't have the same error, so who knows what happened while he was away, Windows updates etc, he mentioned a game install too. Could be the same issue without the error notification, or just some driver conflict while he was away. I believe the card ran fine vertically otherwise.
Jay switched his card into the case and it worked fine.
Yes
@@b.buster. A card that doesn't have issues. They're suggesting it's the orientation on the bad card, that maybe it's some weird pressure that's doing something.
Kinda cool to see someone as knowledgeable as Jay to come across an issue he's stumped with. Sucks for the owner though, I can't imagine how frustrated he must of been this whole time.
That happens all the time, I am playing with computers since like 2004 and still, I see problems which I have no idea what to do with that and those owners think I am magician and I have to fix it in 2 minutes, but it doesn't work like that, sometimes you just have to randomly replace all components because you just don't know where is the problem and sometimes problem will disapear and you have know idea how you fixed it and what was the problem, so you don't know if it's fixed permamently or not. 😀
Nearly 4m subs in and still finding time to stay this "in-touch" with the community. Hats off to you Jayz.
What else was he going to do, finish one of his builds lol
@@thecluckanator lol. That'd be crazy
His name is just Jay. JayZ is/was a rapper. Lmao.
i swear to god when Jay said "imagine we seeing this card go fooosh" the fire animation looks so smooth and clean, who ever is the editor kudos to you.
My thoughts exactly! I was like hell, first viewer build he fixes and it goes up in smoke while filming.
Phil does the editing.
He is that good.
Dont swear to God...thank God
I'd have to believe that it's Phil who's the editor.
Funny the things that stick out in a video. I also stopped for a sec to appreciate how good the flames looked. :D
You should do more of these videos! Everyone has some type of issues and it was great to see you systematically go through all the problems to find out what the cause of the errors are! Keep up the great work!
But he didn't go through the necessary trouble shooting steps especially when knowing that the person have been swapping out components. He didn't rule out the motherboard wiring, the CPU, or the PSU after he noticed that the card worked on his personal test bench. And he can't compare a weaker 3070 card to the 3070 FTW3 when that card draws a little more power. I would've like to have seen the PSU.
I don't remember what psu he even said was in it?
@@spankbuda5760 I agree 100%. He didn't work through this problem at all with a proper scientific or troubleshooting method. *In the video* he states that he has to prove the defect both ways - good gpu fixes it in customer system, but also suspect gpu fails on known good system. He invalidates his theory that the GPU is the problem because it works well in a known good configuration, and then refuses to eliminate further variables or continue with scientific method. Jay found himself in EXACTLY the same place as the customer, because he didn't follow proper troubleshooting steps.
Weird (inconsistent) issues like that typically indicate to a faulty PSU.
Another idea Ive just had: different Mainboards use different CPU-PCIe Lanes for the first PCIe slot, so maybe thats why the GPU works fine in your test-rig. --> reseat the CPU or swap the CPU?
your test bench is horizontal. If there's a bad solder joint on the card it might be orientation specific. So maybe try run his system with the system horizontal instead of vertical.
Also his RAM was in the wrong memory slots which would prevent DOCP from working.
EDIT: Like some other comments have mentioned it might be a PSU issue but I would've expected the GPU undeclock to fix that.
Yep, i was thinking the same as you. That underclock should have reduce spikes on the PSU.
This remember me the strange issue i have with my setup.
My display doesn't turn on after windows initialize the driver. I have to unplug and plug it back in. (Or disable the screen in device manager and re enable it)
My brother has the same display. His work fine. In both mine and his setup. No idea what the issue is...
And yes we swapped the GPUs too but my monitor work in his setup with my GPU.
I have this issue since 2015/2016... I discovered it when i update my GPU from an hd6870 to a gtx1060. I have reinstalled windows multiple times, changed the drives, updated the drivers etc...
It seems a specific issues with that monitor with that GPU (my brother has the 1050ti) and my MB/CPU/ram setup.
Nothing to do.... Well except exchanging the monitors or running a simple script on autorun. (Yes my brother at the end didn't want to swap monitors)
yep Buildzoid you're absolutely right and i noticed and had the same toughts about ram, gpu orientation, and potential psu spikes or power drops issues . also ram in incorrect slots can sometimes create lots of weird issues in some cases .
my guess is, something on the motherboard is pushing against the GPU plastic housing and isnt creating proper contact with the PCIE pins in the slot. I say this because I have a Palit 3080 that due to its design, the frame hits my PCH plastic housing and prevents a fully seated GPU without putting some seriousssssss force on it just to get it too clip in, and even then it still somewhat pushes away from the PCIE slot so i know connection isnt 100%, wondering if something similar is happening here.
DXGI errors are straight up device is no longer visible in device manager (he should check event viewer to see what occurs before that event happens) DXGI errors also have nothing to do with power (which should be ruled out via down clocking anyways) I think to rule out PCIE slot, he should just run it off a riser/extension cable and try both PCIE slots (looks like it has 2)
Yep, I had a VERY similar issue with my MSI Gaming X 3080, worked laying down , but in normal standing position, properly set and secured, would randomly crash. Got one of those GPU support things you can extend to support from the bottom of the case. Haven't had a problem since.
I also do move my setup around a lot and drive it all over the place so maybe that played a part, since it didn't always do this.
As many other commenters have suggested, it may be a faulty PSU. I'd be interested in seeing that PSU tested.
Possibly a ripple that the card doesn't like and causes it to be unstable, or the way the card draws power could be in such a way that the PSU can't sustain whatever demand THAT specific card needs?
This is all speculation on my end, but I hope that you and the viewer can get their system up and running!
550W is "just enough" for system.
Need to be check with OCCT or similar tool for power draw.
"power drops" on 12V, can crash PC suddenly.
I had similar issues and it turned out to be a short in one of the psu cables , when the psu would draw more power under game load it would crash
These troubleshooting videos are the best. Man, just seeing the process of elimination is super inspiring and such a teaching moment
Actually he didn't do the necessary steps in the process of elimination like checking the motherboard bios, the motherboard wiring was properly connected, using a different PSU, or checking the CPU, socket, pins, and the cooler mount and paste.
In my case (DXGI on 2080Ti), it was bad memory stick.
In beginning trying memtest end without errors, but after month of black screens and game crashes with DXGI and TDR errors, finding solutions and triing differnet fixies, i try memtest again, and voila.
Errors in same memory address spaces.
Change memory stick and problems gone away.
Im a bit dissapointed in this teoubleshooting process in this video since it is not a good teoubleshoot
Check 3.3v resistance on the GPU.
Low resistance in the GPU will cause lots of issues to the motherboard.
It can corrupt bios, crash, corrupt ssd etc. Anything that uses 3.3v rail on the motherboard can be hurt by the GPU lowering 3.3v rail down to 300ohms or less.
Other then temps, I'd look at that next.
Best way to diagnose something like that is good old furmark
Out of what I call the big 3 for tech channels, JTC is the only one that does troubleshooting consistently. Having videos that show you troubleshooting a PC helps those who are new at this. Great content as always!
If you want to see troubleshooting videos, I would check Greg Salazars channel out. He usually puts out one a week.
@@TheSjuris When I saw the title while I was scrolling I thought it was another fix or flop video
Ltt is obvs one. Who would you say the third is?
@@YesHumphreyAppleby probably GN.
@@TheSjuris was thinking hardware unboxed / GN
Two things, first I would try a different PSU. There could be a transient spike causing the GPU to crash under load, Gamers Nexus has talked about it a lot lately. The second being trying to update the motherboard BIOS.
Bad transient spikes will completely shut down the system as shown by GN, because it triggers the OCP protection of the PSU. Here the system remains on and functional.
The issue could be PSU or cable related, but it doesn't seem to be caused by massive transients.
@@thelegendaryklobb2879 however it could cause the card VRMs or mosfets to recieve 'dirty power' and if they are strong enough or the PSU isn't strong enough, could cause weird hardware software issues on other hardware. Power is almost ALWAYS my issue on my customer's systems. Whether a bad supply, bad connection, or even just bad power from the wall
i had the same issue... i replaced psu, ram, drivers, changed clock settings, windows etc etc i am as stumped as him. I could relate to his frustration in this video
I thought of the Mobo BIOS too, but it sounds like the user also tried replacing the board. Still wouldn't hurt to check, which I have a feeling Jay did.
@@sleuthchipperson7415 Hardware specs?
We need more videos like this! It's very hard to find videos that you can learn from others experiences.
Greg Salazar has a whole series on this concept of videos. Its called fix or flap
@@Drenixx was gonna say the same thing, good recomendation.
yeah agree some are clickbait and it sucks
Greg Salazar's videos are just awesome and are so much at this point.
I agree with other comments. I have run into similar issues that turned out to be a power problem. I would say the power supply or cables are failing. If you do find the solution please keep us updated.
As other comments I've seen have pointed out, I'd suggest trying a different PSU. The older lesser 3070 would be drawing less power than the FTW version. It would also explain why only some games crash since different games utilize cards differently and can cause much different power draw situations.
on that end surprised no cinabench test. also why no driver booster check? there could be old drivers else where on the system causing a long enough hang for it to only fail on a few games/situations. there's one ive been avoiding: Amd PSP updated 9/16/22 because for whatever reason it causes only my discord and vlc media to error out and crash once its installed. might be something similar.
I'd agree, he could monitor voltages to the card in HWInfo, I'd think if the PSU is not up to snuff you'd see voltage drops.. I had an issue where my machine was crashing, turned out the extension cables I used to dress up my PC were dropping the voltage down to about 11 volts under load on both the 24 pin and PCIe cables. It drove me nuts figuring it out til I saw the drop and the light bulb went off.
I would def try this as I had a bad power supply many years ago. Everyone kept saying that a power supply would not crash the games or the system, but it did. When I played a game it would crash, the hdd even started knocking. Changed power supply - no knocking, all games ran fine.
@@bm5298 Jay reformatted fresh so no driver problem there
@@darkm007 yeah but did he try back dating the drivers after update? I end up renuking windows 8 times before i tried back dating.
More troubleshooting videos please... love this format!
Most everyone's saying it's likely the PSU and I'd agree with that theory. Another approach I would take is matching CPU (or as close as possible) on the test bench since it could be the interaction between the two that causes the problem.
this is very far fetched tbh
I really admire you for taking the time to help your viewer. Props, my fellow old guy!
I definitely want to see an update to this video, don’t disappoint Jay!
Me too. This was a cliffhanger for sure......
@Vercusgames i wouldn't have thought PSU either, but another comment suggested that maybe the card jay tested was older and thus not needing as much power. I don't think they mentioned his PSU in the specs which leads me to believe Jay didn't think it was under supplied.
Hey I had a similar issue with one of my 3070s. The card would work in a vertical mount versus the standard PCIe socket. Note with your experience it worked on your test bench when it was in the vertical position. Assuming you've tested the PSU.
Jay, I had 3 devices now that I have helped people with with almost this exact error. And the error kept getting worse until the system wouldn't even boot (months). All the systems had 30 series cards and rm series Corsair psus. On all of them, swapping the cables to different psu ports or removing extensions fixed the issue. Probably why it works in the other system. Please try to swap the psu
YES
I have a 3080 ftw3 from evga, R9 5900xand a 850rmx and I too had very bad crashes , until I turned off DOCP, and unplugged/plugged in the PSU… and now my games run well
This is amazing to see! I've has a Possessed PC from Hell for 5 years and to see someone like Jay take a random PC and step by step go through his 2 cents on probable causes and then test it! This is REAL content here. I LOVE IT! Please do more problem solving like this (Jay and Other Creators). Thank you again.
I'm surprised he never tested the PSU as it is definitely one of the possibilities. The GPU probably has a higher power draw than the replacement card, even if they are reportedly the same on the box, which is why it worked.
The GPU starves and crashes during transient spikes. Could be the PSU or cables but at this point almost certainly a power delivery issue.
Underclocking was certainly a good test but doesn't eliminate all power related issues.
Power issues cause all sorts of issues in every field of electronics. Noise on radios, corrupted SD cards, random communications failures, Raspberry Pi or computer reboots. It's especially liable to get glossed over because it's not flashy. Until it explodes, that is.
Phil you almost gave me a heart attack with that flame on the card and it's not even my system. That was a beautiful edit.
This was an excellent Troubleshooting 101 type video using a real world scenario. I know this isn't what you want your channel to be, at least to this extent, but I think this is one of your most valuable videos yet for the average PC user to watch.
Agree on both points, the fire edit and video format.
If you want more like that then I highly recommend Gregg Salazar's fix or flop series. That's all he does. Fixes viewers broken gaming PCs.
This is a pretty interesting concept for a video series. On one hand, having a number of different videos that more or less match the search terms that people might look for when they're having computer problems means that you could effectively create a series where you do more or less the same thing and follow the same sensible troubleshooting process. That would make people more effective at troubleshooting their own machines because the biggest problem with folks is that they just try to skip through. Doing the same thing over again and finding different problems in different viewers' machines may help reinforce actually useful troubleshooting methods.
On the other hand, people are idiots and people with tech problems that your videos might help may expect you to help them more in the future. Tricky thing to balance.
theres another creator who does something similar called fix or flop where he does exactly that but the titles arent really in line with an easy troubleshooting catalog
Greg Salazar has already been doing this for quite awhile
these troubleshooting videos are great. we need more of them. excellent work Jay
love these troubleshooting and diagnostic type videos, issues like these are always the biggest pain to pin point to then resolve.
My initial guess, just off given info so far, power issues. Transient spilke from the GPU, but not quite enough to crash the system, or perhaps even a short in the cable to the card. Maybe even a small short on the board itself. Strange multiple cards would do it though.
That was my guess also.
Same im guessing
then it would crash as well with the other GPU he tested out, but it didn't.
Or survive until they're damaged. Put in a new card an all is well until it gets toasted too.
@Paul Frost the viewer's gpu works fine in testing rig so it is not toasted.
Jay if you were looking for some fresh type of content that can do well, I think this video is a good start. Interested in more of these, entertaining and knowledge gain at the same time 🤘
You should check out Greg Salazar's Fix of Flop playlist if you haven't already.
Yep, I agree. Great viewing.👍
@@EricMason0321 yup been following Greg's Playlist too, that's why very interested in Jay's take on it
Agreed.
Hey I have the same issue, glad I'm not the only one.
I would probably try changing PSU... unstable power (especially on higher draw) could lead to all manner of unstable behaviours. And that is one component that is often overlooked when looking at the stability. And it can be possible that this specific piece of card is more prone to crash on power fluctuations, while others are less and just work.
I had a older PSU that kept crashing my whole system for a while so eventually i suspected it may have been PSU so i bought new one (modular this time) and tested it and beside some random startups after shutdown for few days (then it somehow stopped booting up again after shutdown) it works fine so far. This issue definetly seems like PSU or GPU power cable.
I've had parts that just didn't want to cooperate together but worked perfectly fine when separated. The more complex our technology the more variables to consider.
I once saw someone who had a PS3 console that would work with any tv but theirs, and a TV that would work with any PS3 but theirs. No one at Samsung customer service could figure out why that happened
If it’s actually possessed, you should first figure out what type of ghost it is, and Jay is pretty good at that.
The most important step is getting it to tell you its name!
Well what if it’s a Demon instead of a Ghost? Or another type of Spirit even? Or are we assuming Christianity is non-canon in this timeline?
No such thing as ghosts...only demons and angles...
@@kennethporst4359 yes there is there’s a ghost powering my Pc as we speak
Always sucks when you have ghosting issues...
@jay The DXGI error just means the application lost access to it's rendering target, when it says "device" it's not talking about the actual GPU but the virtual device that the DX pipeline targets for rendering. DX / OpenGL rendering does so to a virtual target that the drivers then push to the real screen, it's how we you can alt-tab out of fullscreen and windowed rendering works.
That would only happen if something bad happens to the graphics card itself, my suspicion is the PSU's power delivery isn't clean, different cards have different tolerances and chances are that specific model card is very intolerant of out-of-spec power.
I would try a different PSU. His 3070 might be pulling a little more watts due to the higher clocks. Probably won't do anything but worth a shot. Seems to be the only common denominator in his build
Beat me to it! Absolutely agreed as a troubleshooting step.
That's where I was going too. In my experience weird, random, inexplicable nonsense like this is scarily often a power issue. I always seriously over spec my PSUs in my builds. Power scares me 🙂
My Sapphire Nitro + 6800XT was having DXGI issues as well. Swapped in a Corsair RM850x and bam problem solved. I used HWInfo to log everything while it would crash and noticed literal millisecond voltage drop "gpu disappeared" then the game would crash. Apex was the easiest to get to recreate the problem. I could play Cyberpunk for an hour before it would happen. Drove me crazy!!!!!
@@singlsrvngfrnd yea I have a be quiet dark power 850w psu. Seems to work well with my 6900xt. But I've had thar psu for the past 3 builds. Not sure how many years it has left.
It's almost certainly the PSU, but given Jay under-clocked the GPU and it still had issues I think it might specifically be transients that are causing the issues. IIRC Apex has been said to create transient spikes more frequently, and the 30 series GPUs were notorious for high transients too, so if the PSU dropping is voltage in a transient spike that would certainly explain the behavior shown.
Honestly these types of troubleshooting videos are the best thing ever made. If it wasn't for TH-cam Troubleshooting videos I'd have "bricked" computers because I just wouldn't know what to do.
I would say to try different power channels if he has a modular power supply or do a vertical mount because in your test rig, it worked fine when it was vertically mounted. It might have a problem when it's horizontally mounted.
Can you run mats to test memory ?
It maybe crashing once it fills up enough memory space which may have issues.
My daughter had a very similar situation with a Gigabyte 3070. It would routinely crash when she tried gaming on it, just like the one in the video. We put it in my system to test like Jay did with putting the one in the video in his test bench. It worked fine in my rig for a few days and then started displaying the same behavior. I had tried the Bios and firmware updates just like Jay did. We ended up returning that card and she got an Asus card which has ran flawlessly.
PSU and(or PSU cable problems are a fair suggestion though undervolting should cover some of those. I'd try in this order:
1. Severely underclock the GPU and VRAM under 1 GHz, maybe some undervolt too
2. Check/clean PCI E lanes on the card and pins on the mobo
3. Downgrade mobo's BIOS (for the sake of it)
4. Get a priest
that's what i was thinking, throw msi afterburner or asus gpu tweak III on the machine and start with lowering the power target and maybe the clock just a smidge and see if it persists
The process of troubleshooting PCs is an exercise in logical thinking and working through problems in a structured way, I like your approach.
i would love to see more videos like this! my pc has a new problem every couple of months, so seeing how you fix problems would be awesome. im also everyone's tech support
PSU, potentially the launch card had a lower TPD than the later (factory boosted) FTW3 card, so when put in the viewers system, the launch card could perform ok.
Then when the FTW3 GPU went into Jayz bench, which presumably has a pretty robust PSU, it was able to fully boost.
even with the 120mhz underclock?
@@zalatos possibly he adjusted the clocks, not the power draw. While it might have naturally reduced, depending on how the BIOS is written, it might still ask for the same amount of power when it boosts.
The subscribers card was a 3070 TI FTW3 and he replaced it with a regular 3070 XC3...Definitely a difference in power draw!
@@brucepreston3927 8:34
It's definitely a 3070, not ti, but that doesn't mean power draw isn't the issue.
Like several others have suggested, it could be a power supply or cable issue. Trying a different set of power cables for the GPU would be one place to start. Even though the other card worked, it's probably pulling less power then FTW card. Also as others have said, switching the orientation on the test bench to match how it's installed in the PC and see what happens...
Never ever just switch out power supply cables! Always keep the original cables with the original power supply, very risky otherwise.
@@DarkReturns1 I am a little confused by your comment. There are several companies that make custom replacement cables for a lot of the different brands of power supplies. Now I could understand if you tried to use a cable from a different brand entirely as that may have different pinouts, etc. Even different models from within the same brand. But using a replacement from the same brand and model or using one custom made for your exact power supply shouldn't be an issue.
@@DarkReturns1 you ever heard of cable mods before?
I'd check the PSU. Sounds to me like it's spiking at different points. PSU issues can cause very strange behaviour.
I have an identical pc (except I use a 6700xt), I had the same issue. I updated bios on the board and it fixed the issue instantly.
Weird issues like this can sometimes be caused by a sub-par PSU, might be a transient spike or something that the PSU can't handle properly causing the GPU to not have the voltage it needs and not operating correctly (resulting in a graphics API crash). It would also explain why the GPU works in the bench (better PSU that can handle the spikes) and why another GPU (that might not have as high spikes) could work in the PC.
It might also be worth looking at the event viewer logs to see if there is anything sus going on in there, but given the clean OS install I'd still put my money on the PSU.
Lots of people commenting the PSU, but if you google “apex legends DXGI error device hung” you’ll find endless posts over years of time about this specific crash in Apex. It effects thousands of systems, every config under the sun. A bad PSU wouldn’t explain all of that, there’s no way that many people have faulty PSUs.
Also worth mentioning that I’ve dug through dozens of these posts and videos because I had this problem myself with Apex, never once have I found a real solution including PSU/cable swaps.
Also doesn’t make sense that it effects Apex specifically, so prolifically. Meanwhile myself and basically everyone else who experienced the problem have never had the issue with another game.
Obviously I never found a fix, so I just stopped playing the game for a month or two. Then I went back to it and tried it again and it just worked fine… really weird.
@@charlesbrown4483 DXGI errors can happen for many reasons I'm only suggesting that it is the PSU in this SPECIFIC case because of the behavior of it based on the troubleshooting done. It is possible in your case it is caused by something else.
It makes perfect sense that it affects one game (APEX) more than others. Each game is different, each will put different loads on different hardware component at different frequencies, so if there is a specific issue that happens only under specific conditions then it would only happen if a specific game creates those conditions (which seems to be the case with APEX). A similar thing happened with 3090s which had issues in New World but were perfectly fine in most other games.
It is quite possible that APEX is creating GPU power spikes and the PSU isn't properly handling them which could result in voltage drops causing instability. There are certainly other potential issues but based on the GPU working fine on the bench system, a different GPU working fine in the PC, and the issue persisting between mother board replacements and clean OS installs, I think the PSU is the most likely cause in this particular case.
@@charlesbrown4483 It is also worth mentioning that it isn't that PSUs are faulty, it would likely be that certain PSUs just can't handle the transient spikes created by the GPU to handle the load APEX puts on it. There is a good Gamers Nexus video about transient spikes (which were quite problematic in the 30 series) and GPUs were pulling 2-3 times the normal load. So if the GPU normally pulls 250w of power then a transient spike could pull 750w briefly and not all PSUs would handle that, especially if the total wattage of the PSU was less than that, a 750w PSU would normally be a reasonable PSU for a 250w GPU so it wouldn't surprise me if a large number of PSUs weren't able to handle the spikes properly, especially cheaper PSUs with smaller/cheaper capacitors.
@@AusSkiller Okay yeah I’m with ya, that’s a good theory. Though if we’re talking about transient spikes caused specifically by playing Apex, as you mentioned that’s not a faulty GPU. Which is the conclusion most people are coming to.
So imagine someone has a 750w PSU, they take the advice of the comment section assuming their PSU/cables are bad and just replace the unit with either a new in-box model of the same unit, or just some other 750w unit. In theory there’s no reason that would fix the problem. They’d have just wasted money on a new PSU and should still be getting the crash.
So if this is the case, then the comment section’s advice should be “upgrade to a much higher wattage PSU” instead of “your PSU needs to be replaced.”
Though I gotta be honest I have little faith in any solution for the problem. It’s like the zodiac killer’s puzzle of PC gaming lol.
@@charlesbrown4483 It entirely depends on the 750w PSU, some will be fine with transient spikes up to 1500w, others might struggle with as little as 1000w, and the length of the spike or frequency that they occur can also be factors for if a PSU will be ok. That's part of the problem with the transient spikes most PSUs aren't really rated for them so it is difficult to know which can/can't handle them, and also what kind of transients a component can produce, so it is near impossible to know what you actually need.
My advice for anyone with an issue like this who thinks it might be the PSU is to find a PC repair place (or better yet a PC enthusiast friend) and see if they can test the system with one of their PSUs that is either higher quality or higher wattage to see if that fixes things. If it does then they'll know the PSU is definitely the issue and they can buy that kind of PSU to fix it, if not they might have a different underlying cause.
Agree with others that have already said this. Check power cables/power supply. Since it worked in your test bench, I would say the issue is potentially one where the card is experiencing either a power event or thermal event (test bench is open air so could allow better heat dissipation.
It would be good to use the performance overlay while in the game so you can see the info on the components before it crashes.
So you can see if it some kind of spike compared to your other card that does not crash.
Had almost the same issue as well on an older FX8350 build where it was crashing while loading games, when just the day prior it was running fine. I noticed that the power supply voltages were fluctuating quite a bit. Swapped out the PSU and it has been working perfectly since.
Would love to see more videos like this, modern builds with puzzling issues and how you got around to solving it! Now im eager to see the next video from this and what the actual issue was and how you were able to help solve it for this channel follower! (I'm leaning to the commenters side on this and saying psu/psu cables)
I had the same thing happening with my system. It ended up being the NZXT C650 Power supply went bad. The PCIE portion of the PSU failed. Swapped to a Corsair RM750e and no problems since.
The reason why the GPU works in the test bench is that there the card is not horizontally mounted, which leads to the bad solder ball joint/joints in the affected GDDR6 memory chip to have better contact. The root cause is GPU sag. This is quite typical with these bigger and heavier cards when you have them unsupported for a long time --> the PCB bends and the VRAM chips closest to the PCI-E slot ecentually get bad solder joints due to the mechanical forces. A MATS/MODS test with the card horizontally mounted would reveal the chip with the bad solder joints
holy shit, so ive been having this device removed thing happen, and when I read your comment it hit me, on my case i have a thing to hold up the gpu so u dont get this, but when i upgraded to a 3080ti it was so big i had to remove the thing holding the gpu, and i couldnt figure out how to screw the gpu sag device farther back so i just didnt use it, do u know of any other ways to solve this? maybe ill get some ties or something to hold it up, hopefully Jay reads your comment cause this is likely whats causing it.
as someone who troubleshoots & repairs machines for a living, i really appreciate seeing another professional going through the analytic cadence, and even into the 'this should not have any effect on the fault symptoms, but i got nothing else left to swap out' area :D
i'd live to see more of the cursed pc troubleshooting
I hate that feeling when it feels like you have tons of things left to try and then boom... You suddenly find yourself out of options.
What crazy to me is he completely missed troubleshooting steps that we immediately obvious to me and others in the comments. This would be faulty PSU cables or PSU itself.
i mean, it's also a triage in there. when you consider the symptoms, and get a shortlist of possible causes, most people will ask 'which of these is simplest and fastest to test?' and then work their way up from there. but what is considered simplest by one person is not necessarily same to the other
@@Kholaslittlespot1 I literally wasted months or even years by trying to solve some PC problems, I think it's more like addiction, similar as smoking or playing slot machines, I really have to stop with that and throw all computers to dust bin, becaue I am wasting my best years of life because of that. 😀
@@Pidalin haha I feel you, but I like to think of it more as a passion! In fact, it's helped steer me away from several dangerous addictions in the past. Of course it still needs to stay in balance. I used to have a whole room full of old machines that I eventually gave away to charity... Then again, I still have a habit of picking up old commodore Amigas and ZX spectrums from eBay with the intention of fixing them up, so I'm still pretty bad! xD
The only two things I could think to swap out at this point would be the power supply, because maybe there is something going on with the power rail for the GPU. The card you swapped in wasn't an exact copy of what was already there, so that could explain why yours works in his system, and his works in your system. The other thing I could think of would be try swapping the hard drive itself. You mentioned that there could be an issue with the controller overheating. While I don't believe that to be the case as the system is crashing rather quickly, you may still be onto something with thinking there is an issue with the hard drive itself. Out of the 2 things listed though, my money would be on the power supply.
I remember about a year or so ago that you had a video showing some weird issues with the motherboard bios not liking the video card which caused weird crashing and other issues. Even if this one is on the latest bios, I wonder if a reflash would do anything. After that, my auto mechanic experience kicks in and I'd start looking at physical things, cables and such
2:23 as an IT professional with several certs & almost 20 years under my belt, I appreciate this part. I can't tell you how many people have given me the shocked 😳look when I mention event viewer or looking at logs.
I had Same issue with several games and i had a burned Mosfet for some reason gave it to be fixed and it works now (mine was already a long time after RMA time)
Like other people said, i would try with a differnet psu but before that i would look inside the bios and make sure resizable bar is turned off. Maybe some weird bug in this particular combo of MB / GPU does not like it.
Best video in a long time, Jay at his best trying to solve real issues. Thanks!!
FTW3 Ultra takes a bit more power than a basic 3070, I strongly suspect the PSU. I'd swap the PSU in his system and see if the problem persists.
Had similar issue with my Sapphire 5700xt. Worked fine with other system except the new build. My case was cause of bios. After tested with several old bios version of motherboard which is asrock, I was able to fix the problem.
I actually love these videos, it gives me ideas to fix problems I may or may not have in the future. Thanks for doing this Jay!! 😊👍
From my experience the DXGI issues are from OC on the videocard, happens mostly on EA games
Later Edit: as seeing the card been tested on a different system and working I can only think about the power supply. There was a discussion regarding the peaks of 30 series wich could imply crashes
Ive had this issue. I worked jt down to my oc in msi afterburner. At times it runs fine and then and would randomly crash during an intense fight, sometimes in firing range it would do the same with movement shooting dumies.
This is all in apex using msi 3070ti ventus @ 165/500+ clock. Took it back to stock snd just adjusted fan curve
I'm leaning toward a bad psu or psu connector. If it works in your test bed, then it might be whatever is supplying the GPU. Since virtually every other part has been replaced, then the only two places to look are the GPU and psu. The gpu works in your testbench (which has a beefier PSU). The only option in my mind is his PSU.
Could be. He did not test the psu so Yeah maybe.
I Love this Troubleshooting Vid, Please do more
I think putting one known working ram stick in every ram slot on the motherboard will 100% rule out the ram problem (although the memory channel of the cpu can make things more complex). The second thing I thought of was just clearing the cmos. It is very easy and fast to do, even though that might not be the issue. Same goes for testing ram I mentioned before, fairly easy to do and won't take a lot of time.
One of my old desktop had the same issue but by doing a base clock OC it sort of fix it for a few months before The PUS go kaboom and bring the rest of the parts with it.
@@lowmianxian6479 That really sucks man, I hope you have found your replacing parts for it
First thought was PSU, glad others think so too.
Maybe it's worth checking the PSU. The PSU might not be a high enough wattage or starting to fail. The GPU during these games might be trying to pull too much power for the PSU and when the GPU doesn't get it, it becomes unstable or crashes temporarily to cause these said issues.This would also explain why it works in your system and not there's.
Not entirely sure that's the issue but it'd be the next thing I would consider looking at.
this was really fun to watch, id love to see some type of mini series on this stuff.
also my vote (as someone else said) is on it being a cord/power supply issue, its the only variable that never changed between systems that is literally directly connected with the gpu. both cards you swapped out i believe have lower power draw so maybe the reason they didn't crash and the good cables/psu in your other system ran the card fine.
Jay, move the M.2 to the location ABOVE the video card. Also, the last 3 factors that are remaining as common are the case, SSD, and the POWER SUPPLY!
If the supply is possibly defective or weak on a power rail, it is essentially overloading for a second and tripping a reset internally (I am chucking spitcannonballs here trying to help).
If his supply is borderline for power, or a cheaper brand, it may not be able to handle the load.
Yep that's what I'm thinking too. It's the PSU and/or the cables. That exact video card can't handle that exact PSU.
Agreed PSU. I bet the bench is 1k plus vs the pc with 850w.
This
So - the one item that caught my "eye" was the PSU - Corsair RM850 - I had one of these a few years back and had all the same type of random issues - in the end, replaced everything in the system except for the PSU and still had the same issue, replaced the PSU with the same from Corsair RMA and that one worked for about 6 months and then did the same thing, so got rid of it and went with a EVGA 750 and boom, it was fine - worked like a champ from then on. In fact that system is still running great for my daughter years later.
Yes!
I had a random thing like that as well, replace the PSU, problem solved.
Dang - I returned a Corsair refurbed RM850 to the store (good store, no problems) about two months ago. Was progressively failing earlier and earlier, until system usually could not boot successfully.
I've ran the RM850 for years without issue (It's actually been in 3 different builds so far). you probably "lucked out" and got a bad one.... OR you had dirty power from the power company and it caused the issues. I have never had a Corsair PSU fail for me either, so hearing someone have that kind of problem leads me to believee that the problem was not the PSU, but the power coming into the house, or something else related to it's environment.
@@AKStovall That doesn't mean anything. Just because you never had a problem with a certain brand doesn't mean others haven't. Errors happen all the time in manufacturing. It's usually a small number and probably not reported on much but it does happen, like the melting 4090's, or the incorrectly installed capacitors on motherboards.
Could the power supply be messing with it? Maybe random transient spikes that cause the GPU to freak out?
I just saw this vid, and I was smiling. yeah... sometimes strange things happen. some weeks ago I had to upgrade a comp of a friend of mine... no matter what I did, the comp did stop every time... finally I did smell a burning air ( shortcut smelling)... it was a harddrive with a broken power connector. A peace of plastic between the poles was gone.. It's one of those things you never thought such things could happen, but it kept me busy for at least 40 hours. I don't know if you already solved the problem, but I am sure you did..
make this a series please, I love this kind of content
tune in to Fix or Flop - Greg Salazar for a great trouble-shooting series.
His hair is gray enough allready 🤣🤣
Fairly surprised, but pleased you did a piece on this--seems to be a rather 'large' issue that no one has been talking about.
I'd be curious to see what happens if you try hooking up another power supply to his system. Could be that his card being the FTW3 is tripping a bad power supply whereas your card is not pulling as much power since it's an xc3 3070.
Thank you! It would've been nice to see the PSU. And also, he should've checked the motherboard wiring connections and the CPU to rule that out.
I had a similar issue; my PC would crash only during gaming (randomly). I replaced everything but the CPU. I then one day ran Prime95 longer that did before (because I fell asleep and walked off) and it gave me an error on just one core about 18+- hours later. Had AMD replace my 5600x CPU that fixed it. I normally stress test all my hardware 24 hours minus the GPU/CPU for only 30 minutes and 2 hours of Haven on loop. I am sharing this hoping it helps others with isolation of problems. Please run Prime95 for 24 hours part of their burn in on new builds/overclocks.
It's a very interesting issue to run into. It's just that card that's causing issues in that particular system. I had seen another comment where someone recommended checking the power supply, I would try another GPU that has a higher power draw and see what happens. Also, checking eventviewer will show you what kind of issues there are and give you a tiny bit more insight on what's going on (very rare occasion that it helps). Any other steps would be very involved and time-consuming, but you never know what could happen until you try them out! Great job on the troubleshooting, I'd pay good money for your service any day!
As many here, I vote for the PSU because I had some strange issues recently with this part and it looks like this computer had the same. I don't know the specs of your backup GPU but it may ask a little bit less power or it is asking for it a more stable way. Anyway, this is why everything works fine when you switched it.
My own experience : until mid-2021, I had a EVGA SuperNova 650-P2 PSU with a i5-4590, 32 GB DDR3, a motherboard and a Gigabyte 1070 TI. I wanted to upgrade it to a Ryzen 3700x on a Gigabyte B550 AORUS Pro and more RAM. The GPU would stay because shortages at the time.
The EVGA 650-P2 was recent because the Corsair I had died one night, and we can safely say that this is high-quality stuff. I NEVER spare money on the PSU, better safe than sorry.
But the new parts were never able to go to POST. I sent back first the CPU (because LEDs on the motherboard pointed at it) and got a replacement who didn't work better, so I replug back my old parts and sent the new motherboard back to buy an another brand... but the issue was still here (with a difference : I could start it, but not more than one or two minutes before the screen goes black).
Strangely enough, I plugged back the old parts and they worked fine again. 2 motherboards, 2 CPUs... so it couldn't be anything else than the PSU at this point. I asked a friend to lend me his own unit (sorry for the disassembly moment my friend...) and tested my theory : it was indeed the PSU. The new parts weren't supposed to ask more power than before, but my EVGA PSU wasn't unable to provide it. RMAed, the replacement unit was fine.
Event decent PSUs can be faulty, sometimes in strange ways.
Thanks for sharing this content, Jay. I always struggle with troubleshooting but you have given me more ideas to look at next time I troubleshoot a PC!
I have seen the same issue before, the pci-e x16 lanes of the cpu might be broken/unstable. it also happend to be on a ryzen 5000 CPU .
I would be trying that GPU with a different PSU with different cables, and comparing pins on the motherboard and GPU. Seems more likely to me that it'd be a perfect storm of tiny problems that combine into an issue that isn't replicable with any other hardware combination.
even though the motherboard was RMA'd?
@@zalatos yes, even in spite of an RMA’ed motherboard. For one, it’s not guaranteed that the user was absolutely correct on the course of events, and for another, it’s just a look at the contacts. Seems wildly dense to not tames few minutes to take a closer look at them even if that’s not the problem.
Great video Jay. I always appreciate seeing an experts thought process as they troubleshoot some weird issue.
Love these troubleshooting videos! Even if we don't see the resolution, seeing the process is really helpful!
I had the exact same problem. I did all the things that you did, and was never able to resolve it. I finally just ended up buying a new card, and all has been fine since.
Squad is typically a fairly stable game, too... But random crashes with no discernable cause was linked to my PSU getting tired, especially on games that could spike the framerate extremely high.
The 550 watt PSU might not be able to handle the current spikes from the GPU.
The FTW3 might be pulling just enough power over the base 3070 that it can't keep up.
Try a second PSU running just the GPU to see if it'll stay running.
I had to do that for my 3090 while I waited for a replacement PSU to come in, lol.
Edit: misheard it. It's an 850w PSU, but still sounds like a power issue.
I didn't even hear it was a 550 watt. Yeah, that's clearly the problem.
Corsair RM850 PSU is what Jay says, not 550.
i think it is a 850watt
I think you hit it on the head. A simple look on the web shows that the 3070 ftw3 should use a minimum 650 watt PSU. Considering how much RGB the user has in their system, they're definitely having a power problem where the GPU is trying to suddenly draw huge amounts of power and is unable to and probably causing the power to cut off to the GPU, which in turn would make the system think the card has been removed.
Yes, I misheard the first time, it is an 850 according to the closed captions. New models making me hear things, lol... I still think it's a power issue, though. May be a bad solder job on the card, though.
You should do more of these ''Computer from Hell''. I'm riveted at the edge of my seat and dying of laughter at the same time.
Super interesting! I want more
Some tips:
1) Some issues can be solved by using different video output ports on the card i.e. using another DP or HDMI output. (try them all because some ports share silicon, think of it as usb slots sharing same USB hubs)
2) Trying using different PCIe slot and one that has a 8x electrical slot instead of x16 and/or lowering to PCIe 3.0 (or down to 2.0) might help with signaling, if signaling is the issue that corrupts. (if this solution helps then a workaround might also be lowering the motherboard PCIe buss speed, sometimes 99MHz helps when normal is 100MHz, or go even lower)
3) Perhaps it is a physical problem, If there are bad soldering you can perhaps inspect the card and maybe bake the card.
4) Lastly a long shot, have a look if there are no extra ATX standoffs creating this issue.
interesting suggestions.
Being in the market of pc repair for over 17 years, I have seen this issue with many Nvidia and AMD gpus. And I have always found swapping out power supplies to be the best and last guess in troubleshooting DX and DXGI errors. Sometimes Pci-e power cables just go bad. And it's frustrating.