GPU woes

TheValeyard

Well I'm back again

Was playing Cyberpunk2077 earlier, when the game crashed to desktop and since then any game I try to play crashes back to desktop.

I've DDU drivers and reinstalled. I removed the ram and GPU and reslotted it. Remove the motherboard battery, check Bios, tried defaults. Tried earlier GPU drivers. I'm at a loss, is the card or my PSU dying?

It will load into windows fine. I can play older games like Alien Isolation or 2D games such as 'Graveyard Keeper', but any graphic intensive game just crashes back to Windows, even older ones such as Hitman 2 or Assassins Creed Odyssey.

Cannot see any other problems.

Current setup, 3090, 32GB ram, i7-6900k, Win10, m.2 boot drive, 3 x ssd drives, AE-5 soundcard, corsair 860i psu

K.O.Kiki

Just to double-check but are you using two dedicated 8-pin power connectors to the PSU?

Spear

If anyone can't see that image correctly, right click and choose view image (or your browser variation of it). Seems it doesn't display right on here.

On the GPU side, what are the temps like? More stressful games will understandably ramp the temp more. Are the fans still running and ramping up?

_CreeD_

TheValeyard wrote: »

Remove the motherboard battery, check Bios, tried defaults.

It's kind've implied but not 100% - Just to be clear do you mean to hard-reset the BIOS (as in via the Jumper) with the CMOS battery out and system unplugged? Pulling the battery on it's own is not enough. If you haven't gone this far then absolutely do, it can resolve all sorts of weird issues. I had a hard crash on mine a day or 2 after installing a new 3090 and it wouldn't even boot to BIOS, monitor came on but zero actually onscreen, a hard reset resolved it.

Also as mentioned above load up a diagnostics suite and look for obvious culprits like temperature spikes, power draw etc. for both GPU and CPU. Hwinfo with logging enabled is a good choice as it will write your states every few seconds to a CSV so it can be evaluated post crash, https://support.lumion.com/hc/en-us/articles/360021829573-How-do-you-provide-a-log-file-from-HWiNFO- .

ironictoaster

This is so bizarre.

JayzTwoCents just posted with the exact same issue as you.

Hopefully you don't have to RMA it.

https://youtu.be/3puInnaG5ok

TheValeyard

Cheers all, will try suggestions and report back

Samuel T. Cogley

Can you cap it's boost behaviour? Maybe it's right on the edge like they were before.

TheValeyard

K.O.Kiki wrote: »

Just to double-check but are you using two dedicated 8-pin power connectors to the PSU?

Im using the dedicated pins that came with the PSU and theres two 8-pins joined together and seperate 8 pin for 3rd slot on the GPU

Spear wrote: »

If anyone can't see that image correctly, right click and choose view image (or your browser variation of it). Seems it doesn't display right on here.

On the GPU side, what are the temps like? More stressful games will understandably ramp the temp more. Are the fans still running and ramping up?

Very sure its the rights pins, temps were mid 60s. Fans are running smooth and ramping up when needed.

_CreeD_ wrote: »

It's kind've implied but not 100% - Just to be clear do you mean to hard-reset the BIOS (as in via the Jumper) with the CMOS battery out and system unplugged? Pulling the battery on it's own is not enough. If you haven't gone this far then absolutely do, it can resolve all sorts of weird issues. I had a hard crash on mine a day or 2 after installing a new 3090 and it wouldn't even boot to BIOS, monitor came on but zero actually onscreen, a hard reset resolved it.

Also as mentioned above load up a diagnostics suite and look for obvious culprits like temperature spikes, power draw etc. for both GPU and CPU. Hwinfo with logging enabled is a good choice as it will write your states every few seconds to a CSV so it can be evaluated post crash, https://support.lumion.com/hc/en-us/articles/360021829573-How-do-you-provide-a-log-file-from-HWiNFO- .

Ill try the link, and maybe do the battery again just to be sure.

ironictoaster wrote: »

This is so bizarre.

JayzTwoCents just posted with the exact same issue as you.

Hopefully you don't have to RMA it.

https://youtu.be/3puInnaG5ok

Will watch

Samuel T. Cogley wrote: »

Can you cap it's boost behaviour? Maybe it's right on the edge like they were before.

cap its boost behaviour?

Samuel T. Cogley

TheValeyard wrote: »

cap its boost behaviour?

Instead of letting it go to say 1988Mhz cap it at 1950Mhz.

TheValeyard

Samuel T. Cogley wrote: »

Instead of letting it go to say 1988Mhz cap it at 1950Mhz.

Ah yeah, tried that. Went down by 10, but I might try more and see.

TheValeyard

Weird thing is, I can get into the main menus of these games, just loading a level, every game crashes back to Windows.

Squidgy Black

TheValeyard wrote: »

Weird thing is, I can get into the main menus of these games, just loading a level, every game crashes back to Windows.

That just means it's crashing under load, when your gpu usage starts to increase. Could possibly be the power supply.

I'd say try the PCIe cables in a different order, and then maybe try undervolt it. Which model 3090 is it?

TheValeyard

Squidgy Black wrote: »

That just means it's crashing under load, when your gpu usage starts to increase. Could possibly be the power supply.

I'd say try the PCIe cables in a different order, and then maybe try undervolt it. Which model 3090 is it?

Tried changing cable order and then did some undervolting. No difference.

It's the MSI 3090 Gaming X Trio

It's worked grand for the past month, no problems on Valhalla, Legion or other games.

GHOST MGG

TheValeyard wrote: »

Tried changing cable order and then did some undervolting. No difference.

It's the MSI 3090 Gaming X Trio

It's worked grand for the past month, no problems on Valhalla, Legion or other games.

jayz2cents gpu model was the 3080 gaming x trio...i wonder if their is something in that..could be a dying card..watch his video if you havent already..he had to RMA his card

EDIT: after watching that video again it was a 2080ti that was borked sorry if i confused you..he replaced it with a working 3080 gaming x trio..thanks to the lads who pointed that out in the thread..im going to go hide in the corner now:-0

SickBoy

I'm thinking your PSU isn't up to the job and has had enough.
I've heard that the 3090 alone can spike to 800w
Do you have an overclock on the CPU? If you do it's just putting further load on a PSU that's already creaking.

TheValeyard

GHOST MGG wrote: »

jayz2cents gpu model was the 3080 gaming x trio...i wonder if their is something in that..could be a dying card..watch his video if you havent already..he had to RMA his card

Contacted msi about Rma.

My last role of the dice is a complete reinstall of Windows, see if there is some software conflict.

TheValeyard

SickBoy wrote: »

I'm thinking your PSU isn't up to the job and has had enough.
I've heard that the 3090 alone can spike to 800w
Do you have an overclock on the CPU? If you do it's just putting further load on a PSU that's already creaking.

Turned off the overclocks including xmp

weisses

GHOST MGG wrote: »

jayz2cents gpu model was the 3080 gaming x trio...i wonder if their is something in that..could be a dying card..watch his video if you havent already..he had to RMA his card

I'm positive that jays faulty card was a 2080 ti

SickBoy

weisses wrote: »

I'm positive that jays faulty card was a 2080 ti

Your right, he mentioned that it would almost be impossible to get a replacement now.

ED E

Squidgy Black wrote: »

That just means it's crashing under load, when your gpu usage starts to increase. Could possibly be the power supply.

Nah, disagree. Most games load the card during intro and to an extent in menu.

Weird thing is, I can get into the main menus of these games, just loading a level, every game crashes back to Windows.

VRAM.


When you load the game you load very little into VRAM, when you load a level the game pulls all the textures it can fit into VRAM to start rendering.

Wonda-Boy

Well I have the Palit 3090 Gamerock and I am using 3 dedicated 8 pins....it would be a hell of alot more tidy to use only two by splitting one of them (as each cable has two 8 pin tops if you get me) but its deffo not recommended.

I have an 850w Gold Rated PSu.

No issues here as of yet only have it a day but I was gaming like a mad eejit last night for hrs, and if it helps the temps of the card @ stock was late 50c and I am using the standard fan curve in the Thundermaster suite.

I would deffo be leaning towards the PSu here, if you want I have a Coolermaster 1000w PSu here if you want to try it....just for process of elimination fella? If it were me I would hate to be having trouble with my new card...so if it helps you, you more then welcome to try it and see.

TheValeyard

Wonda-Boy wrote: »

Well I have the Palit 3090 Gamerock and I am using 3 dedicated 8 pins....it would be a hell of alot more tidy to use only two by splitting one of them (as each cable has two 8 pin tops if you get me) but its deffo not recommended.

I have an 850w Gold Rated PSu.

No issues here as of yet only have it a day but I was gaming like a mad eejit last night for hrs, and if it helps the temps of the card @ stock was late 50c and I am using the standard fan curve in the Thundermaster suite.

I would deffo be leaning towards the PSu here, if you want I have a Coolermaster 1000w PSu here if you want to try it....just for process of elimination fella? If it were me I would hate to be having trouble with my new card...so if it helps you, you more then welcome to try it and see.

Appreciate the offer dude. Make is hopefully giving me a lend of his PSU, see if thats the issue. Not looking forward to all that lovely cable basterding!

_CreeD_

I came back to suggest what you already mentioned above, don't rule out a software issue with windows and do a reset/reinstall just to be sure.
Also try disconnecting all SATA drives except what you need for testing. Weird as it sounds another issue I ran into a few years back, an old SSD that on the surface worked fine but on a new build was causing random but frequent crashing and eventually corrupted the windows install (which was on a different drive, a symptom of the repeated hard crashes ). On the surface it looked like a low level cpu/ram/psu load failure, I guess the drive was somehow causing feedback into the board.

TheValeyard

Did a clean install of windows. Clean install of GPU drivers afterwards. Same probs. Looks like I'll have to play the PSU or GPU game.

Spear

I can't find it mentioned yet, but does the Windows Event log capture anything from this?

Wonda-Boy

Can you get a lend of another high tier card like a 3080 or even a 3090?

Where are you based....?

TheValeyard

Wonda-Boy wrote: »

Can you get a lend of another high tier card like a 3080 or even a 3090?

Where are you based....?

In Dublin, have a possible RTX card to use all going well.

Dunfyy

You went overboard taking out battery etc
You might have to reinstall bios that you download from other computer. First thing you check is memory one at a time different slots you can install beep speaker to find out how many fault beeps. You have. You also should have tried to go into bios to see if there's problems
Most intel computers don't need graphics card as they can use interpreted unless the have f on cpu so you can take it off board
Some windows 10 updates are crashing ssd hard drive so you may need format and reinstall new you also need to check power supply and wiring if you have friend with computer test parts on their machine try resetting bios when computer starts

Adamant it
hasn't some in depth youtube repair videos but they are a bit long

TitianGerm

Dunfyy wrote: »

You went overboard taking out battery etc
You might have to reinstall bios that you download from other computer. First thing you check is memory one at a time different slots you can install beep speaker to find out how many fault beeps. You have. You also should have tried to go into bios to see if there's problems
Most intel computers don't need graphics card as they can use interpreted unless the have f on cpu so you can take it off board
Some windows 10 updates are crashing ssd hard drive so you may need format and reinstall new you also need to check power supply and wiring if you have friend with computer test parts on their machine try resetting bios when computer starts

Adamant it
hasn't some in depth youtube repair videos but they are a bit long

Good to see you back again Bobby.

Wonda-Boy

Dunfyy wrote: »

You went overboard taking out battery etc
You might have to reinstall bios that you download from other computer. First thing you check is memory one at a time different slots you can install beep speaker to find out how many fault beeps. You have. You also should have tried to go into bios to see if there's problems
Most intel computers don't need graphics card as they can use interpreted unless the have f on cpu so you can take it off board
Some windows 10 updates are crashing ssd hard drive so you may need format and reinstall new you also need to check power supply and wiring if you have friend with computer test parts on their machine try resetting bios when computer starts

Adamant it
hasn't some in depth youtube repair videos but they are a bit long

HELLO EVERYBODY>THIS IS YOUR DAILY DOSE OF THE INTERNET>

Bungusbeefcake

What titles are you crashing on? And what hard drive are they running off? Have you thought that it could be your hard drive dying? or a version of your DX has a corrupted sector?

Also, agree with the windows updates f***ing everything. Maybe not hard drive specifically but could have been linked. I'm running the PC from an M.2 Drive.
My PC updated on it's own over like 20 mins last week and before them it was running fine at a 5.2GHz OC, and XMP running RAM at 3200. After update it restarted, crashed....crashed again, wouldn't load boot screen, wouldn't run recovery without crashing. I had to do a CMOS reset (which is luckily easy with the MSI MEG Z490 Unify as there's a reset CMOS button). This was the only thing that then allowed me to restart windows. Then I had to manually reset any of my OCs and it's fine now.

Samuel T. Cogley

Tried turning the hardware scheduling off? Just a random thought.

TheValeyard

Bungusbeefcake wrote: »

What titles are you crashing on? And what hard drive are they running off? Have you thought that it could be your hard drive dying? or a version of your DX has a corrupted sector?

Also, agree with the windows updates f***ing everything. Maybe not hard drive specifically but could have been linked. I'm running the PC from an M.2 Drive.
My PC updated on it's own over like 20 mins last week and before them it was running fine at a 5.2GHz OC, and XMP running RAM at 3200. After update it restarted, crashed....crashed again, wouldn't load boot screen, wouldn't run recovery without crashing. I had to do a CMOS reset (which is luckily easy with the MSI MEG Z490 Unify as there's a reset CMOS button). This was the only thing that then allowed me to restart windows. Then I had to manually reset any of my OCs and it's fine now.

Crashing on any game post 2013. Did cmos reset. Will check HDD and SDD again.

TheValeyard

Samuel T. Cogley wrote: »

Tried turning the hardware scheduling off? Just a random thought.

Will have a look. Cheers.

TheValeyard

By the way, just want to say thanks very much for the responses so far. It's very much appreciated.

Been onto MSI about RMA but they are telling me to contact the retailer not them.

ED E

TheValeyard wrote: »

Been onto MSI about RMA but they are telling me to contact the retailer not them.

Theyre right. Your rights are with the retailer.

ED E

TheValeyard wrote: »

Crashing on any game post 2013. Did cmos reset. Will check HDD and SDD again.

We've nothing to point to storage medium. Wild goose chase. Same for CMOS resets unless it'd fix some magical XMP issues but again, system is normally stable so that doesn't make sense.



Run Furmark, if its stable that says the GPU cores are fine and the PSU is able to push ample juice to keep you up.

TheValeyard

ED E wrote: »

We've nothing to point to storage medium. Wild goose chase. Same for CMOS resets unless it'd fix some magical XMP issues but again, system is normally stable so that doesn't make sense.



Run Furmark, if its stable that says the GPU cores are fine and the PSU is able to push ample juice to keep you up.

Yeah Furmark crashed after less than two seconds

Although increasing the memory clock and core clock it runs for a lot longer. I'm very confused.

jelem

You need to take heed of here re bugs -- https://en.wikipedia.org/wiki/Cyberpunk_2077
i have found to trust my set up and look elsewhere first.

ED E

It took a lot of digging to find out which '90 you had, when posting a troubleshooting thread its helpful to be specific.

(Its an MSI Trio X 3090 as far as I can tell)



What was the end result of the MLCC v POSCAP drama?

Although increasing the memory clock and core clock it runs for a lot longer. I'm very confused.

Are you overriding the boost by manually clocking? This could prevent the card boosting over the cards limit and crashing it.


1. Run Afterburner
2. Set Power Limit to 90%
3. See if you can game.

TheValeyard

ED E wrote: »

It took a lot of digging to find out which '90 you had, when posting a troubleshooting thread its helpful to be specific.

(Its an MSI Trio X 3090 as far as I can tell)



What was the end result of the MLCC v POSCAP drama?







Are you overriding the boost by manually clocking? This could prevent the card boosting over the cards limit and crashing it.


1. Run Afterburner
2. Set Power Limit to 90%
3. See if you can game.

Will try when home.

TheValeyard

ED E wrote: »

It took a lot of digging to find out which '90 you had, when posting a troubleshooting thread its helpful to be specific.

(Its an MSI Trio X 3090 as far as I can tell)



What was the end result of the MLCC v POSCAP drama?







Are you overriding the boost by manually clocking? This could prevent the card boosting over the cards limit and crashing it.


1. Run Afterburner
2. Set Power Limit to 90%
3. See if you can game.

Tried reduucing power limit to 90% then 85%, still crashing.

Apologies, its the MSI 3090 Trio X.

the MLCC V POSCAP? As in Furmark, i can run it on Preset: 720 (getting 320fps) Anything else crashes within 2-8 seconds.

Im hoping to have a 1200w within couple of days from a mate, see if thats a a fix of sorts. Same Corsair brand and type, just more power to it. If not, ive been onto the retailer I bought it from, but yet to hear back a reply from CCL.

I just have my doubts there's nothing else at play here. Seems odd a brand new GPU can work fine for a month, then just crash out, yet still load older games like 'Graveyard Keeper' or run windows, Netflix etc, no problem, yet crash out under a higher load.

Redfox25

With the higher capacity PSU it should help you rule out your own PSU as the culprit.
Assuming its a decent quality 1200W, some of the cheaper units arent great and might be worse than what you have already, rated for 1200W and delivering about 800W.

Good luck with it, hope you get it sorted.

TheValeyard

Redfox25 wrote: »

With the higher capacity PSU it should help you rule out your own PSU as the culprit.
Assuming its a decent quality 1200W, some of the cheaper units arent great and might be worse than what you have already, rated for 1200W and delivering about 800W.

Good luck with it, hope you get it sorted.

Cheers, it's a Corsair PSU, should be good quality.

TheValeyard

Bungusbeefcake wrote: »

What titles are you crashing on? And what hard drive are they running off? Have you thought that it could be your hard drive dying? or a version of your DX has a corrupted sector?

Also, agree with the windows updates f***ing everything. Maybe not hard drive specifically but could have been linked. I'm running the PC from an M.2 Drive.
My PC updated on it's own over like 20 mins last week and before them it was running fine at a 5.2GHz OC, and XMP running RAM at 3200. After update it restarted, crashed....crashed again, wouldn't load boot screen, wouldn't run recovery without crashing. I had to do a CMOS reset (which is luckily easy with the MSI MEG Z490 Unify as there's a reset CMOS button). This was the only thing that then allowed me to restart windows. Then I had to manually reset any of my OCs and it's fine now.

I finally got around to tinkering more with this idea. I disabled all hard drives except for my boot SSD drive. Literally plugged the other four drives out.
Furmark has been running grand for about 30 minutes now, zero crashes, loading steam and newer game onto boot drive. See if it loads up.

You maybe onto something here!!

Edit: yeah no joy, still crashes.

Wonda-Boy

Mate of mine went with a MSI 3090 trio X aswell the same day I got my 3090. He had alot of trouble with his out of the box, he specifically had trouble with the PCI lanes set to either AUTO or 4x in the bios......might be worth a look and see if can change the settings in there for process of elimination.

Samuel T. Cogley

Tbh you've a faulty card there at this stage. Same experience I had, loads of thing looking like they could be a fix but nothing actually sorting it.

TheValeyard

Wonda-Boy wrote: »

Mate of mine went with a MSI 3090 trio X aswell the same day I got my 3090. He had alot of trouble with his out of the box, he specifically had trouble with the PCI lanes set to either AUTO or 4x in the bios......might be worth a look and see if can change the settings in there for process of elimination.

Switched it to Gen3 from auto. Same problem. Cheers though.

TheValeyard

Samuel T. Cogley wrote: »

Tbh you've a faulty card there at this stage. Same experience I had, loads of thing looking like they could be a fix but nothing actually sorting it.

Cheers, probably looking that way alright. Although I've yet to hear back from ccl, been well over a week now, must just be that busy time of year and Christmas time off.

N64

See if you can get an alternative PSU to test (sorry if already tried). My mate had a Corsair PSU cause complete system shutdowns and boot cycling. Apparently there's some QC issues with them for models produced this year. Source: https://www.reddit.com/r/hardware/comments/kh3yg5/faulty_corsair_hx1200_hx1200i_psus/

Dunfyy

Cyberpunk crashes they updated some of it to stop crashes but it still crashes