PC restarts under heavy load after replacing PSU

[harm9963]

I learn the Hard Way .

https://blog.tripplite.com/pure-sine-wave-vs-modified-sine-wave-explained​

 

[nomdeplume]

If bought new in 2022, the APC Back-UPS 1000VA/600W UPS, would be plenty if its being plugged directly into an outlet, rather than through a power strip (unless it's faulty of course). The new PSU shouldn't cause it to go into safety/security mode. Something is borked, particularly if the rig is functional without the UPS.

We all know how that goes. On the ground reporting displays feasibly solid reasoning discrediting official figures. Not worth chasing our tails over when the solution lies in moving beyond.

In any number of other cases the PSA would be worth it's weight in gold.

 

[RJARRRPCGP]

If bought new in 2022, the APC Back-UPS 1000VA/600W UPS, would be plenty if its being plugged directly into an outlet, rather than through a power strip (unless it's faulty of course). The new PSU shouldn't cause it to go into safety/security mode. Something is borked, particularly if the rig is functional without the UPS.

Safety mode should keep it off and force you to unplug and replug in.

 

[Calenhad]

Yeah, but OP's clicks and shuts off while using it, so I was referring to that. The computer didn't just reboot, the power supply clicked, shut off, then clicked back on and the system came back. I just think it helps limit the possible problems if we know that whatever it is actually forces the power supply to shut off fully (on its own), then turn back on. I wasn't indicating that the clicking in itself is a problem.

The PSU will click if the motherboard instructed it to shut down to. For example if the motherboard OVP/OCP/... tripped. I had a motherboard that did this for a couple months before it died entirely.

If bought new in 2022, the APC Back-UPS 1000VA/600W UPS, would be plenty if its being plugged directly into an outlet, rather than through a power strip (unless it's faulty of course). The new PSU shouldn't cause it to go into safety/security mode. Something is borked, particularly if the rig is functional without the UPS.

APC claims an estimated battery life of 3-5 years for that unit. But I am unable to find any in-rush tolerance. They specify it as 600W maximum period. There is an overload indicator on the display though. So you should be able to tell if that is your issue.
Otherwise it is possible that you have a motherboard issue for example, see above.

 

[ynneB]

I think this is the best immediate course of action you can take. Let us know how it goes!

Good info! So, that UPS is much better and would likely not be over-drawn by your system under normal conditions. The "bad battery" may be partially to blame though. I haven't used a UPS at home, but we have some at work that will just beep continuously and then shut off the output after a while when the battery fails. So it's possible that the battery in your UPS has been failing, which has been causing your issues. If you can run all your games and testing on mains power, that definitely points to the UPS.

Also, as for inrush current, it may very well exceed the limits of the UPS, so that could be another issue (smaller, but also annoying due to the beeping). The inrush is a short event, but I'm not sure how long the UPS would give an alarm from that event. I don't think you'd be exceeding the 600W under any normal conditions with your computer, but the inrush may exceed it briefly. Just for reference, the inrush happens when power is first applied and all the filter-capacitors are charged up (think of them sort of like fast acting batteries). It takes a huge spike of current to charge those capacitors, which then hold your DC power stable through any dips or fluctuations later. It's a normal and unavoidable situation, but I suspect that there's more filter capacitance on your new supply than there was on the old one and this gives you a higher spike. Between the higher spike and likely degradation of your UPS battery (which are usually replaceable, by the way), this is likely the two problems you're dealing with (1. higher spike causes the beeping, 2. battery dying may cause the UPS output to fail under load).

The PSU will click if the motherboard instructed it to shut down to. For example if the motherboard OVP/OCP/... tripped. I had a motherboard that did this for a couple months before it died entirely.

APC claims an estimated battery life of 3-5 years for that unit. But I am unable to find any in-rush tolerance. They specify it as 600W maximum period. There is an overload indicator on the display though. So you should be able to tell if that is your issue.
Otherwise it is possible that you have a motherboard issue for example, see above.

Hello all, hope you are having a good day

Using PC while connected to mains did not help, system restarted in less than 2 hours of playing games (1 hour of Forza then 5 mins into a BFV game it went ahead and restarted). So suppose you can rule out UPS as well.

I've done some research; apparently despite the Seasonic PRIME PSU series being an A-Tier PSU it has issues with RTX 3080/3090 and 6800 XT GPUs as stated from cultists.network:

Seasonic PRIME based units experience shutdowns with RTX3080/3090 (and possibly RX6900 XT) GPUs. The cause is not the OCP tripping but a PSU design flaw as evident by the PSU not latching off on shutdown and 1000W+ models being affected too. Doesn’t manifest in 100% cases as it’s also dependent on motherboard model and GPU OC. Seasonic provides a replacement 24-pin ATX cable to fix this via support. Appears to be fixed post 2021 although there’s no official confirmation from Seasonic.

Confusing part is, I don't use any one of those GPUs by a long shot but the bit where it says its dependent on motherboard model and GPU OC has me skeptical. Are they stating that it can have problems on specific mobos and GPU OCs while not using those listed GPUs? (My GPU is an OC'd variant from Gigabyte)

Not sure what else to do now. To my knowledge it seems that it's either mobo and PSU (although not entirely sure whether it is PSU thanks to the insight of the wonderful people here) considering this never, ever happens in light tasks or idling, only when during demanding tasks as I've mentioned above.

Also, is it ok if I plug the PC back into the UPS since the issue is happening regardless?

Thank you for your time

 

[maxfly]

Next, you really need to eliminate the GPU. Do you have an old card you can test with? If not you're going to need to figure out a way to either go get it tested or get a loaner. It might be the MB but it's far less likely to be the MB. And it's much easier to test for a bad GPU than to source another z170 mb. Your PSU is likely fine being that the issue persists between both your old and new PSUs correct?

Running off the UPS is gtg now that you're sure that's not the culprit.

 

[ynneB]

Next, you really need to eliminate the GPU. Do you have an old card you can test with? If not you're going to need to figure out a way to either go get it tested or get a loaner. It might be the MB but it's far less likely to be the MB. And it's much easier to test for a bad GPU than to source another z170 mb. Your PSU is likely fine being that the issue persists between both your old and new PSUs correct?

Running off the UPS is gtg now that you're sure that's not the culprit.

Hello, the issue on the old PSU was entirely different at least in my opinion, it would happen during same circumstance (i.e playing demanding games) EXCEPT on the old PSU my monitors lose output (No Signal to be specific), I can hear in-game audio for a few seconds then i don't hear anything, my keyboard stays lit up, my system continues running (Fans and all) and I literally can't do anything aside from shutting off the PC via the PSU switch. I also check the event log each time and I either get EventID 14 (Kindly check attached picture, it's not my screenshot but this used to be the exact error I used to get) and another EventID relating to dwm.exe (I think it said something along the lines of dwm.exe - Unknown Hard Error. (I cannot find a screenshot for this on the internet but I'm 90% certain that was the error description). All rails on PSU were off too, 11.7v on 12v rails and same -0.3 or -0.4v increments lower on the remainder of the rails too

The moment I replaced the old PSU with the one I'm currently using, PC abruptly restarts during demanding task. No black screen symptom as mentioned earlier, no shut downs, just restarts. No events logged either aside from "The System has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed or lost power unexpectedly." as I mentioned on initial post.

I will again try asking around if any of my colleagues have a spare GPU so I could test, it is very difficult especially in my country where they are already hard to come by as they are very expensive purchases (even entry level, mid range cards, more so current and especially old motherboards) so instead people play on console or laptop.

Thanks for continued assistance

Hello, another update:

I have run prime95 for 3 hours and and 1 run of y-cruncher to test my CPU/RAM oc config with no errors.

Now, as mentioned by maxfly to possibly rule out GPU, upon some research I have run Unigine Superposition benchmark mode on 1080p medium preset. (With my OC config)

During the very 1st run of the benchmark, the moment it transitions to the 15th scene my PC restarts like usual.

Now, I decide to try this again except with stock clocks on both CPU and RAM. I run Superposition again and it finished benchmarking successfully 2 times and on the 3rd run it restarted like before.

Not only that, i have come across another symptom, after restart, my keyboard stops working entirely unless I disconnect and reconnect it from the USB cable. This happened on 1 of the 2 times it restarted after running Superposition benchmark.

What could this mean?

P.S I've asked people I know about borrowing a spare GPU and nobody has one; let alone a z170 motherboard so very clueless on what to do next.

Thank you

Edit: To mention that I also ran prime95 at OC frequency to verify CPU/RAM stability for 3 hours like how I've run it at stock clocks yesterday.

 

[Bobaganoosh]

Well, more information is always good, but the intermittent problems are painful sometimes. Glad to hear you're trying to isolate.

Given the new info, it seems that we can rule out the UPS as the primary issue (based on what you described before there may be a battery issue, but that doesn't seem to be the main problem you're dealing with).

One thing about the CPU clocks difference, when the CPU is more capable, it puts more strain on the video card (by moving more workload to it), so now I would agree that the GPU is the most likely issue. Being that it usually needs to be loaded heavily for the fault to happen, and it's worse when the CPU is overclocked, I think another test would be helpful.

Being you have superposition, can you run the 4k Optimized test? That should put a heavy load on the GPU and relatively light load on the CPU. With the 6600k, I'm curious if 1080p medium still puts a heavy load on the CPU.

What I'm curious about is if your GPU is having an overcurrent event under heavy use or an over-temperature event that causes it to shut-down. As others said, the motherboard will give a signal to reboot the power supply, and being you've now tested two supplies, I think it's probably not the supply. With the temperature concern, this would be worse with the overclocked CPU as it will load the GPU harder, but also add more heat to the system. The way that you described your first crashes (on the first supply) happening, it sounds like something your card doesn't have a monitored sensor for could have overheated (memory, VRM, etc.) and that triggered the restart. This has happened with a lot of GPUs over the years. With yours working for a long time then suddenly having an issue, I'm wondering if your thermal interface pads/material have gone bad. So testing another GPU would help isolate this to the GPU, but I understand it's not easy to just borrow a GPU sometimes.

If you're comfortable with it, you could try to order some new pads/thermal interface material for your GPU (some use various thickness pads, some use putty, etc.) and do a re-paste. This would require disassembling your card, replacing all the thermal paste, pads, putty, etc. and then putting it back together. Only do this if you are comfortable with it (or find someone who is). You'd have to do some research to find out what Gigabyte used for pad thicknesses and then try to order replacements (from Gelid, Fujipoly or similar). Alternatively you can buy a new card...but often with old video cards or laptops, you can rejuvenate them with a fresh application of thermal interface materials. This is assuming it is a thermal issue and not a power failure of some kind on the card. The fact that it takes so long to happen and is worse with higher temps in the case leads me to think it's more thermal than power, but I just want to be clear that I couldn't be 100% sure of any diagnosis online.

Another very simple thing I would try before doing anything complicated is to just reseat the GPU. As you heat-cycle the system (every time you get things warm over days, months, years, etc.) things can shift and flex a little bit and although the GPU is supposed to be locked into the PCIe slot, sometimes things can shift just a little. So I'd remove the GPU, check the slot for any damage, then re-install it, make sure it's fully seated, then make sure the monitor and power cables are fully re-installed and test the system again.

 

[agent_x007]

Is your GTX 980 Ti reference design (blower) with OC on top ?

 

[evernessince]

I've done some research; apparently despite the Seasonic PRIME PSU series being an A-Tier PSU it has issues with RTX 3080/3090 and 6800 XT GPUs as stated from cultists.network:

Seasonic PRIME based units experience shutdowns with RTX3080/3090 (and possibly RX6900 XT) GPUs. The cause is not the OCP tripping but a PSU design flaw as evident by the PSU not latching off on shutdown and 1000W+ models being affected too. Doesn’t manifest in 100% cases as it’s also dependent on motherboard model and GPU OC. Seasonic provides a replacement 24-pin ATX cable to fix this via support. Appears to be fixed post 2021 although there’s no official confirmation from Seasonic.

Confusing part is, I don't use any one of those GPUs by a long shot but the bit where it says its dependent on motherboard model and GPU OC has me skeptical. Are they stating that it can have problems on specific mobos and GPU OCs while not using those listed GPUs? (My GPU is an OC'd variant from Gigabyte)

Not sure what else to do now. To my knowledge it seems that it's either mobo and PSU (although not entirely sure whether it is PSU thanks to the insight of the wonderful people here) considering this never, ever happens in light tasks or idling, only when during demanding tasks as I've mentioned above.

Unfortunately that article by Cultists.network is incorrect and you are probably the 30th person I've see quote them without realizing it. Completely understandable given they keep that content up.

JonnyGuru (former PSU reviewer and current PSU engineer at Corsair) stated that the issue stems from the Nvidia 3000 series giving noise feedback on the +12v sense line.

You can view that here: https://forums.tomshardware.com/thr...tform-high-current-tripping-resolved.3734867/

So it's an issue on Nvidia's end, one that they never cared to fix. Similar to how they haven't issued a recall for 12VHPWR equipped cards despite the newer 12v6x2 update to the connector shortly after.

In addition, no Prime unit made 2018 or later are affected as Seasonic updated their cables to fix the issue even though they were not at fault to begin with. Existing prime owners had the option to have replacement cables shipped to them if they had a 2017 and earlier model.

Also, is it ok if I plug the PC back into the UPS since the issue is happening regardless?

Thank you for your time

Yes, given the PC is still crashing without the UPS you can hook it back up. It's possible that the UPS and another part is causing issues but given the likilihood of that being extremely low we can stow that concern for now.

You could try running off the iGPU and see if you get any crashing during gaming. It won't test your PSU but it will give you an idea of if the graphics card is at fault. Given that you've run CPU stress tests for 3 hours and a full MemTest we know those aren't the cause of the issue. The possible culprits are either the GPU, Motherboard, and PSU in that order.

I also check the event log each time and I either get EventID 14 (Kindly check attached picture, it's not my screenshot but this used to be the exact error I used to get) and another EventID relating to dwm.exe (I think it said something along the lines of dwm.exe - Unknown Hard Error. (I cannot find a screenshot for this on the internet but I'm 90% certain that was the error description).

That appears to be a graphics driver crash.

Do you still have warranty on this newer PSU? Can you test with another PSU? This is troubleshooting is taking way too long - let's simplify it - if the problems started with Seasonic unit (and they were known to be defective at around that time), then try another one and return this one if you can.

The OP's issue started before he had his current Seasonic PSU. The Prime series was not known to be defective, please read above in my comment for details including quotes from Corsair's head PSU engineer.

 

[ynneB]

Well, more information is always good, but the intermittent problems are painful sometimes. Glad to hear you're trying to isolate.

Given the new info, it seems that we can rule out the UPS as the primary issue (based on what you described before there may be a battery issue, but that doesn't seem to be the main problem you're dealing with).

One thing about the CPU clocks difference, when the CPU is more capable, it puts more strain on the video card (by moving more workload to it), so now I would agree that the GPU is the most likely issue. Being that it usually needs to be loaded heavily for the fault to happen, and it's worse when the CPU is overclocked, I think another test would be helpful.

Being you have superposition, can you run the 4k Optimized test? That should put a heavy load on the GPU and relatively light load on the CPU. With the 6600k, I'm curious if 1080p medium still puts a heavy load on the CPU.

What I'm curious about is if your GPU is having an overcurrent event under heavy use or an over-temperature event that causes it to shut-down. As others said, the motherboard will give a signal to reboot the power supply, and being you've now tested two supplies, I think it's probably not the supply. With the temperature concern, this would be worse with the overclocked CPU as it will load the GPU harder, but also add more heat to the system. The way that you described your first crashes (on the first supply) happening, it sounds like something your card doesn't have a monitored sensor for could have overheated (memory, VRM, etc.) and that triggered the restart. This has happened with a lot of GPUs over the years. With yours working for a long time then suddenly having an issue, I'm wondering if your thermal interface pads/material have gone bad. So testing another GPU would help isolate this to the GPU, but I understand it's not easy to just borrow a GPU sometimes.

If you're comfortable with it, you could try to order some new pads/thermal interface material for your GPU (some use various thickness pads, some use putty, etc.) and do a re-paste. This would require disassembling your card, replacing all the thermal paste, pads, putty, etc. and then putting it back together. Only do this if you are comfortable with it (or find someone who is). You'd have to do some research to find out what Gigabyte used for pad thicknesses and then try to order replacements (from Gelid, Fujipoly or similar). Alternatively you can buy a new card...but often with old video cards or laptops, you can rejuvenate them with a fresh application of thermal interface materials. This is assuming it is a thermal issue and not a power failure of some kind on the card. The fact that it takes so long to happen and is worse with higher temps in the case leads me to think it's more thermal than power, but I just want to be clear that I couldn't be 100% sure of any diagnosis online.

Another very simple thing I would try before doing anything complicated is to just reseat the GPU. As you heat-cycle the system (every time you get things warm over days, months, years, etc.) things can shift and flex a little bit and although the GPU is supposed to be locked into the PCIe slot, sometimes things can shift just a little. So I'd remove the GPU, check the slot for any damage, then re-install it, make sure it's fully seated, then make sure the monitor and power cables are fully re-installed and test the system again.

Hello, firstly I want to say thank you all so much for assisting me all the way through ! It brought much peace of mind where I would otherwise feel helpless if not for this forum. The fact you take time out of your day to help out a novice like me is very much appreciated


I watch many repasting and servicing GPU videos. Earlier I watched one from Greg Salazar where he fixed a PC that restarted the moment Furmark was run. Turns out repasting GPU fixed it completely (The TIM was not applied onto the die). I'd be willing to do the same thing if we are 101% sure it's GPU though it'll take a couple of weeks to get pads and putty as I'm very sure those aren't available locally in Maldives (They charge $22 for updating windows, resetting bios and cleaning pc with a vacuum lol).


Also, aside from benchmarking on iGPU and reseating GPU shall I try to install GPU on the 2nd PCI-E slot (I attached picture of the motherboard) and benchmark to see if problem persists? I also suspect there's a bit of GPU sag and I'll make sure to send you a picture of that so you'll be to better determine whether it is or not.

Unfortunately that article by Cultists.network is incorrect and you are probably the 30th person I've see quote them without realizing it. Completely understandable given they keep that content up.

JonnyGuru (former PSU reviewer and current PSU engineer at Corsair) stated that the issue stems from the Nvidia 3000 series giving noise feedback on the +12v sense line.

You can view that here: https://forums.tomshardware.com/thr...tform-high-current-tripping-resolved.3734867/

So it's an issue on Nvidia's end, one that they never cared to fix. Similar to how they haven't issued a recall for 12VHPWR equipped cards despite the newer 12v6x2 update to the connector shortly after.

In addition, no Prime unit made 2018 or later are affected as Seasonic updated their cables to fix the issue even though they were not at fault to begin with. Existing prime owners had the option to have replacement cables shipped to them if they had a 2017 and earlier model.



Yes, given the PC is still crashing without the UPS you can hook it back up. It's possible that the UPS and another part is causing issues but given the likilihood of that being extremely low we can stow that concern for now.

You could try running off the iGPU and see if you get any crashing during gaming. It won't test your PSU but it will give you an idea of if the graphics card is at fault. Given that you've run CPU stress tests for 3 hours and a full MemTest we know those aren't the cause of the issue. The possible culprits are either the GPU, Motherboard, and PSU in that order.




That appears to be a graphics driver crash.



The OP's issue started before he had his current Seasonic PSU. The Prime series was not known to be defective, please read above in my comment for details including quotes from Corsair's head PSU engineer.

Hello, again this is very valuable information especially regarding PSU so thank you so much for your assistance, hopefully we'll be able to rule out PSU after tests are done. (Do check above on what I'm planning on doing later)

Is your GTX 980 Ti reference design (blower) with OC on top ?

Nope, its triple fan with a very beefy cooler (please check picture attached).

Thank you all kindly

 

[Bobaganoosh]

Short of testing your gpu in another computer or testing another similar one in your computer (with the restart issue following the card), it's difficult to be 100% certain. You do have clear evidence of Nvidia driver errors and the restarts happen when the GPU is under heavy load so it definitely points that way, but I wouldn't say 100% sure.

I'd be surprised if there's anything wrong with your actual first Pcie slot, but reseating the GPU can sometimes fix a bad connection and it's a super simple thing to check off the list, especially if there's a lot of sag and it may have shifted over time.

 

[ynneB]

Short of testing your gpu in another computer or testing another similar one in your computer (with the restart issue following the card), it's difficult to be 100% certain. You do have clear evidence of Nvidia driver errors and the restarts happen when the GPU is under heavy load so it definitely points that way, but I wouldn't say 100% sure.

I'd be surprised if there's anything wrong with your actual first Pcie slot, but reseating the GPU can sometimes fix a bad connection and it's a super simple thing to check off the list, especially if there's a lot of sag and it may have shifted over time.

Hello, thanks for sticking with me once again! I may have found a pattern that can possibly determine the cause of the problem. I don't know what it could be exactly but here are my findings:

Another thing I forgot to point out, I have a dual monitor setup: BenQ XL2411 connected via DVI and an ASUS VS238-H connected via HDMI. This could be important down the line but I don't know.

Anyway, I reseated GPU and swapped to the extra spare GPU power cable that Seasonic psu had in the box.

After that, As you have suggested, I ran Superposition on 4k and PC restarted into 2nd scene. This is with both monitors connected to GPU.

Later, I run same benchmark but with DVI monitor unplugged 10 times. No restart.

At that point I decided to unplug the HDMI monitor and run same benchmark 10 times. No restart.

Here's the confusing bit; when I was done testing HDMI monitor by benchmarking superposition 10 times as earlier, the moment I decide to plug in the DVI monitor back into the GPU (I didn't even make contact with the DVI port of the GPU), I either hit the hdmi cable or some other cable in the system and that triggered a restart. Note that I have tried to lightly tug all the cables on the back of my PC after the restart (obviously not the psu power cable) and I could not replicate this.

Again, after that, I plug both monitors to GPU again to see if it restarts while running Superposition and it does so after 3rd benchmark run.

I will see if I can replicate the restarts this on single monitor except while playing games in a bit. This is all very confusing, as you can see. I assume GPU can't handle running two monitors at once 7 years into its life cycle. Don't know for sure.

Thank you

EDIT: Forgot to mention I have also run superposition with same config on iGPU with GPU removed and ran it 3 times with no restarts. Just lots of info to share and its overwhelming at times, apologies

 

[evernessince]

Also, aside from benchmarking on iGPU and reseating GPU shall I try to install GPU on the 2nd PCI-E slot (I attached picture of the motherboard) and benchmark to see if problem persists? I also suspect there's a bit of GPU sag and I'll make sure to send you a picture of that so you'll be to better determine whether it is or not.

I think that's a great idea.

Hello, thanks for sticking with me once again! I may have found a pattern that can possibly determine the cause of the problem. I don't know what it could be exactly but here are my findings:

Another thing I forgot to point out, I have a dual monitor setup: BenQ XL2411 connected via DVI and an ASUS VS238-H connected via HDMI. This could be important down the line but I don't know.

Anyway, I reseated GPU and swapped to the extra spare GPU power cable that Seasonic psu had in the box.

After that, As you have suggested, I ran Superposition on 4k and PC restarted into 2nd scene. This is with both monitors connected to GPU.

Later, I run same benchmark but with DVI monitor unplugged 10 times. No restart.

At that point I decided to unplug the HDMI monitor and run same benchmark 10 times. No restart.

Here's the confusing bit; when I was done testing HDMI monitor by benchmarking superposition 10 times as earlier, the moment I decide to plug in the DVI monitor back into the GPU (I didn't even make contact with the DVI port of the GPU), I either hit the hdmi cable or some other cable in the system and that triggered a restart. Note that I have tried to lightly tug all the cables on the back of my PC after the restart (obviously not the psu power cable) and I could not replicate this.

Again, after that, I plug both monitors to GPU again to see if it restarts while running Superposition and it does so after 3rd benchmark run.

I will see if I can replicate the restarts this on single monitor except while playing games in a bit. This is all very confusing, as you can see. I assume GPU can't handle running two monitors at once 7 years into its life cycle. Don't know for sure.

Thank you

Assuming you don't see crashing in games with a single monitor config, that rules out the monitors and monitor cables themselves as the source of the issue. If the issue is a monitor / cable, it would have still have issues in a single config. Seems to indicate a GPU issue although I suppose it could also be a driver bug. After you are done testing that, perhaps you can try a different driver version? You have a 980 Ti so perhaps you could try something pre-4000 series and then pre-3000 / 2000 series?

 

[ynneB]

I think that's a great idea.



Assuming you don't see crashing in games with a single monitor config, that rules out the monitors and monitor cables themselves as the source of the issue. If the issue is a monitor / cable, it would have still have issues in a single config. Seems to indicate a GPU issue although I suppose it could also be a driver bug. After you are done testing that, perhaps you can try a different driver version? You have a 980 Ti so perhaps you could try something pre-4000 series and then pre-3000 / 2000 series?

Sorry, I quoted you but for some reason forum only showed your username on the quote message.

Considering the restarts happen less than 2 days into gaming, i will only be able to say something after that.

I'm on 546.17 drivers, which is what I updated to when I had the black screen issue on old PSU. Never updated it since. Prior to that I was running 496.76, will rollback if restarts persist on single config.

Update: On single monitor config, not even 10 minutes into Forza, it restarts. I also heard faint clicks in the PC seconds before the restart, don't know where it was coming from exactly.. Maybe hard drive? Its about 7 years old now and I've read that they can cause restarts too when they start failing. confusing considering we have checked health info on crystaldiskmark.

 

[evernessince]

Sorry, I quoted you but for some reason forum only showed your username on the quote message.

Considering the restarts happen less than 2 days into gaming, i will only be able to say something after that.

I'm on 546.17 drivers, which is what I updated to when I had the black screen issue on old PSU. Never updated it since. Prior to that I was running 496.76, will rollback if restarts persist on single config.

Update: On single monitor config, not even 10 minutes into Forza, it restarts. I also heard faint clicks in the PC seconds before the restart, don't know where it was coming from exactly.. Maybe hard drive? Its about 7 years old now and I've read that they can cause restarts too when they start failing. confusing considering we have checked health info on crystaldiskmark.

Could always try unplugging the HDD to see if that fixes the issue. Might be worthwhile to disconnect any USB devices that aren't needed as well should you have any.

 

[thesmokingman]

Did you ever unplug from the UPS and test?

 

[ynneB]

Could always try unplugging the HDD to see if that fixes the issue. Might be worthwhile to disconnect any USB devices that aren't needed as well should you have any.

The hard drive is the only storage device I have on my pc unfortunately.. I've also unplugged everything aside from essentials (keyboard, mouse) during tests.

Did you ever unplug from the UPS and test?

Yes, I have mentioned so on Post #30

 

[thesmokingman]

The hard drive is the only storage device I have on my pc unfortunately.. I've also unplugged everything aside from essentials (keyboard, mouse) during tests.


Yes, I have mentioned so on Post #30

Ah I see. At this point go buy another psu to test. Your symptoms still scream psu to me.

 

[evernessince]

The hard drive is the only storage device I have on my pc unfortunately.. I've also unplugged everything aside from essentials (keyboard, mouse) during tests.


Yes, I have mentioned so on Post #30

Hm, that makes things a tad more difficult. Have you already tried removing the GPU and just running off the iGPU? I don't quite remember if that was mentioned earlier.

 

[Bobaganoosh]

Can you try to take some good pictures of the top of your GPU with it still installed in the system? (the opposite side of the fans...where the circuit board is). Especially close to the DVI and HDMI connectors. Also, take a look at the monitor cables and connectors to see if any of the pins look bent or anything looks out of place. The idea would be to rule out a short circuit that only happens after things heat up and flex. Tugging might not show this sort of issue, but wiggling the cables or sometimes pushing on them a little may. Same goes for other cables. It still sounds like GPU is the main culprit, but I'm just not sure why or if that's just because it's the largest/hottest load in the system.

 

[ynneB]

Hm, that makes things a tad more difficult. Have you already tried removing the GPU and just running off the iGPU? I don't quite remember if that was mentioned earlier.

hello, i've mentioned so that I did run PC off iGPU after removing GPU and ran same 4k superposition tests with no issues. you probably missed it as I edited the message later, sorry about that !

Can you try to take some good pictures of the top of your GPU with it still installed in the system? (the opposite side of the fans...where the circuit board is). Especially close to the DVI and HDMI connectors. Also, take a look at the monitor cables and connectors to see if any of the pins look bent or anything looks out of place. The idea would be to rule out a short circuit that only happens after things heat up and flex. Tugging might not show this sort of issue, but wiggling the cables or sometimes pushing on them a little may. Same goes for other cables. It still sounds like GPU is the main culprit, but I'm just not sure why or if that's just because it's the largest/hottest load in the system.

I'll take pics of the circuit board area but wouldn't I have to take off the gpu backplate so that you'd be able to see it? it comes with one installed.
I've only been able to check HDMI cables and they seemed OK (initially thought one pin on the right side on each HDMI port was missing then I looked it up and saw its like that cause of HDMI design). DVI cable couldn't really take a look as it was late yesterday.

I've also noticed that the gpu ports are insanely dusty and ive read lots of dust build up can cause problems too. its a weekend here and I can't think of a place to get 90-100% IPA right now. Are there any other alternatives?

EDIT: once I'm home, shall I try rolling back gpu driver? I have not updated gpu drivers ever since installing new PSU and updating BIOS.

 

[Bobaganoosh]

I'll take pics of the circuit board area but wouldn't I have to take off the gpu backplate so that you'd be able to see it? it comes with one installed.
I've only been able to check HDMI cables and they seemed OK (initially thought one pin on the right side on each HDMI port was missing then I looked it up and saw its like that cause of HDMI design). DVI cable couldn't really take a look as it was late yesterday.

I've also noticed that the gpu ports are insanely dusty and ive read lots of dust build up can cause problems too. its a weekend here and I can't think of a place to get 90-100% IPA right now. Are there any other alternatives?

EDIT: once I'm home, shall I try rolling back gpu driver? I have not updated gpu drivers ever since installing new PSU and updating BIOS.

Ah, I couldn't tell if your card had a backplate or not. Pictures of your actual system wouldn't hurt though. Areas to focus on for pictures would be cable connections, PCIe slot area (while installed, just sort of showing the inside of the case with the camera pointed that way). If you can see in the gap between your GPU's PCB and the backplate, a picture of that gap might actually be helpful.

As for dust, you can try to just blow it out (with the system off). You don't want to blow shop-style compressed air (like 80-100PSI you'd have for putting air in tires) at your fans, but I don't know if you have a blower of some kind (like a shop vac that you can run a nozzle on the output, an air mattress pump, cheap "electronics blowers" like this, or a can of dust-off like this (just don't hold it upside down lol), etc.). You just make sure the PC is off and unplugged from AC, then you blow out all the connections and dust you can. You can blow the air off fan blades, but hold the fan with your hand or something so it doesn't spin. The risk is if you get the fan spinning fast enough, it can generate voltage and if high enough damage things. Usually parts are protected against this, but as a general rule of thumb it's a bad thing to do. Don't freak out if the fans move or anything, it'd have to spin faster than it's normal 100% speed to even pose a risk.

While you can clean PCBs with IPA, you have to be careful about what you're scrubbing the PCB with and not damage any of the surface mount components. Usually if you can blow the dust off without touching anything it's safer all around.

 

[ynneB]

Ah, I couldn't tell if your card had a backplate or not. Pictures of your actual system wouldn't hurt though. Areas to focus on for pictures would be cable connections, PCIe slot area (while installed, just sort of showing the inside of the case with the camera pointed that way). If you can see in the gap between your GPU's PCB and the backplate, a picture of that gap might actually be helpful.

As for dust, you can try to just blow it out (with the system off). You don't want to blow shop-style compressed air (like 80-100PSI you'd have for putting air in tires) at your fans, but I don't know if you have a blower of some kind (like a shop vac that you can run a nozzle on the output, an air mattress pump, cheap "electronics blowers" like this, or a can of dust-off like this (just don't hold it upside down lol), etc.). You just make sure the PC is off and unplugged from AC, then you blow out all the connections and dust you can. You can blow the air off fan blades, but hold the fan with your hand or something so it doesn't spin. The risk is if you get the fan spinning fast enough, it can generate voltage and if high enough damage things. Usually parts are protected against this, but as a general rule of thumb it's a bad thing to do. Don't freak out if the fans move or anything, it'd have to spin faster than it's normal 100% speed to even pose a risk.

While you can clean PCBs with IPA, you have to be careful about what you're scrubbing the PCB with and not damage any of the surface mount components. Usually if you can blow the dust off without touching anything it's safer all around.

Hello, here are some pictures, i didn't have much access to a blower style unit to effectively get all dust out of the ports and connectors but i've done the best i could. pins on the cables seem ok, though after taking pics of the GPU i notice sag as i said before.

Anyway, after all this i plugged pc like i usually do, rolled back drivers to 496.76 and ran superposition once. Restarted right before 3rd scene.

At this point i am just going to hand over pc to a local service center sometime this week and explicitly tell them to test the pc while swapping GPUs cause i'm pretty sure that's the case at this point.

Thanks for the help all.

 

[ShrimpBrime]

Hello. I've read the 1st post and your last post. Nothing in between.

From the first post, my thoughts.

The black screen with audio still playing is directly the GPU dropped the video driver. You could use keys Shift-Ctrl-Windows key- B key all at once to reset the driver.

This happens from tired graphics units, bad overclocks and driver instability from one of the first two. Perhaps even from under-volt on very rare occasions.

The issue I have is the system in its entirety is lack for a better word, Old.

The PSU would need to be tested on hardware that's new. Cause now without diagnostics have switched a part, we'd be chasing our tail after this. I'd go back to testing with the old PSU and find a definitive answer. But then we are talking about old hardware.

From any lack of experience, gotta say that system memory can cause similar issues to the original back screen. LPX is as bottom bin as it goes.... All this knowledge without the PC is a difficult diagnostic online in a forum. The PSU would not have been my first guess.

GL!