PC Shutdown on GPU use

[noteffy]

Hi,

 

I have had this PC for more than 2 years now, and had no issues until a week or two ago.

 

i5-4460 - CPU

MSI - H97M-E35 - MOTHERBOARD

Kingston HyperX Fury 1x8GB DDR3-1866 - RAM

Samsung EVO SSD 120GB - STORAGE (Windows)

Samsung EVO SSD 240GB - STORAGE (Games)

Western Digital 1TB HDD 7200RPM - STORAGE (video editing etc...)

Sapphire Radeon r9 290 4GB Tri-X - GPU

Fractal Design Define S (no window) - CASE

XFX - TS 650W  80+ Gold - POWER SUPPLY

 

Recently, while playing CSGO, once or twice my PC shutdown without any warning, no bluescreen, nothing. Only to start back up in a second on its own (this is normal as this happened months ago when we had electricity shortage couple of times)

 

I've since downloaded Fortnite and unless I set the settings to the lowest settings, the PC will shutdown before I even enter the game.

I have played a few games on low settings without shutting down.

 

I have tried GPU stress test (FurMark) which crashes a couple of times just as I started the stress test and it went to full screen.

 

I followed this by checking various forums, which had no clear answer what is at fault.

I proceeded with uninstalling and installing GPU drivers, installing malwarebytes, WhoCrashed and then closed all the programs that were currently running (after PC restart)

The FurMark got to 95-100% done and the PC shutdown (I tried this the last 2 times). In total using FurMark 3 times.

 

I have no way of trying another GPU or PSU.

WhoCrashed says No valid crash dumps have been found on your computer

 

Is there a way to tell which is the culprint?

 

Thank you,

and sorry for the wall of text, wanted to give as much information as I had.

 

EDIT:

forgot to add, temperatures were looking all fine and normal.

although I am not sure how to see PSU temperatures.

 

 

[emosun]

23 minutes ago, noteffy said:

although I am not sure how to see PSU temperatures.

You don't , power supplies are not digitally connected to the mainboard.

What was the gpu temp? My r9 290 that I had back in the day was essentially the most unstable POS I've ever owned. They run hot/loud/and power hungry.

The one thing that made my 290 work was locking the fan at 80% when I played games , and when not playing games I had to underclock the gpu 50% when doing anything 2d.

so maybe try undercloking yours to 85% power and lock the fan at a very high speed and see what that gives you.

[Xander Cousins]

Tbh sounds like a PSu problem if it is crashing when playing games and doing high GPU tasks, It's as if the PSU cannot supply the required power anymore.... I don't know of any way of checking if it's your PSU unless you have a secondary one to replace it in the RIG and try with 

[noteffy]

22 minutes ago, emosun said:

You don't , power supplies are not digitally connected to the mainboard.

What was the gpu temp? My r9 290 that I had back in the day was essentially the most unstable POS I've ever owned. They run hot/loud/and power hungry.

The one thing that made my 290 work was locking the fan at 80% when I played games , and when not playing games I had to underclock the gpu 50% when doing anything 2d.

so maybe try undercloking yours to 85% power and lock the fan at a very high speed and see what that gives you.

Idle 47. on the full use it was 80-90 if I remember correctly.

I have had no issues with the GPU for 2 years, so I guess I was lucky

Not sure how to underclock, never done it.

 

9 minutes ago, Alex Colson said:

Tbh sounds like a PSu problem if it is crashing when playing games and doing high GPU tasks, It's as if the PSU cannot supply the required power anymore.... I don't know of any way of checking if it's your PSU unless you have a secondary one to replace it in the RIG and try with 

That's what I thought, from the things I have read about similar situations, more of then than not, the solved cases, when its GPU's fault they get blue screen or some sort of warning/message and not just straight up PC crash/shutdown

 

 

[Xander Cousins]

Should be able to use MSIAfterburner to overclock and underclock your GPU, done it once when my old car acted up, That may help your problem but I am pretty sure it is a PSU supply that is at fault if it kicks of then Reboots a second later with no error message.

[emosun]

4 minutes ago, noteffy said:

on the full use it was 80-90 if I remember correctly.

should be more like 75 under load , if it's 90 the gpu won't break at that temp , but it can very easily have errors/crash at that temp

5 minutes ago, noteffy said:

Not sure how to underclock, never done it.

in the display settings it's under global settings/overdrive. move the power limit and clock speed down some.

you can leave the target gpu temp high so it won't thermal throttle , but manually lock the gpu fan speed with something like speedfan. lock it at 80% and see what you get.

[raphidy]

There's no real way to test which is the part that cause the problem without switching the parts physically. Can't you borrow parts from a friend? I would test the PSU first since you have problem on full load.

[noteffy]

1 hour ago, Alex Colson said:

Should be able to use MSIAfterburner to overclock and underclock your GPU, done it once when my old car acted up, That may help your problem but I am pretty sure it is a PSU supply that is at fault if it kicks of then Reboots a second later with no error message.

 

1 hour ago, emosun said:

should be more like 75 under load , if it's 90 the gpu won't break at that temp , but it can very easily have errors/crash at that temp

in the display settings it's under global settings/overdrive. move the power limit and clock speed down some.

you can leave the target gpu temp high so it won't thermal throttle , but manually lock the gpu fan speed with something like speedfan. lock it at 80% and see what you get.

I used msi aftrburner to underclock core clock from 1000 to 900. not sure if thats what i was suppose to do, but i did. I also used MSI afterburner to set fan speed to 80%. Sure it did keep the temperatures, which i played 1 game to test, but the PC sounded like a jet...

Fortnite did not crash.

Tried FurMark, as before it crashed pretty much couple of seconds after I started the test.

 

32 minutes ago, raphidy said:

There's no real way to test which is the part that cause the problem without switching the parts physically. Can't you borrow parts from a friend? I would test the PSU first since you have problem on full load.

That is the problem, the closest friend that might be able to borrow me a PSU is about 6 hours drive.

The friends where I currently live do not have PC's. notebooks and macbooks are what they use...

 

 

[emosun]

2 minutes ago, noteffy said:

Fortnite did not crash.

Tried FurMark, as before it crashed pretty much couple of seconds after I started the test.

thats progress

 

underclock it more to see if furmark stabilizes. I know it's loud and slow , but you're just testing to see if these changes effect everything.

[noteffy]

29 minutes ago, emosun said:

thats progress

 

underclock it more to see if furmark stabilizes. I know it's loud and slow , but you're just testing to see if these changes effect everything.

Okay.

The test i did before was 900. and do keep in mind that i did these tests once. if you need me to do more tell me, its just i don't think its that good crashing the PC so many times

 

900 - crashed at 'warmup' on furmark. 

800 - crashed at last second of benchmark (95-100%) temp was 69

700 - crashed at last second of benchmark (95-100%) temp was 67-68

At this point I was like it just keeps crashing...I'll test 500 and be done with it. guess what.

500 - did not crash (was not looking at temps)

600 - did not crash (was not looking at temps)

650 - crashed at last second of benchmark (95-100%) this time i checked temp it was 68.

 

Fan always on 80%

clock speed as above.

memory clock was never changed as you didn't tell me to, stayed at 1300

[emosun]

in that case try underclocking it using the amd settings. should be under global setting then overdrive.

it might be a bit more stable using it's own drivers to run at slower speeds.

 

But essentially it looks like the card isn't getting enough power possibly and underclocking it is reducing it's power draw with afterburner. if you try with overdrive instead it might reduced the cards power in a better way than afterburner.

[noteffy]

2 minutes ago, emosun said:

in that case try underclocking it using the amd settings. should be under global setting then overdrive.

it might be a bit more stable using it's own drivers to run at slower speeds.

 

But essentially it looks like the card isn't getting enough power possibly and underclocking it is reducing it's power draw with afterburner. if you try with overdrive instead it might reduced the cards power in a better way than afterburner.

I go to display settings and cannot see what you're after.

They look like this. https://gyazo.com/6bab5fc68fe0f07b7b2ea1d01659f18c

[emosun]

at the bottom it says gaming / then go to overdrive tab 

[noteffy]

3 minutes ago, emosun said:

at the bottom it says gaming / then go to overdrive tab 

Not there...

https://gyazo.com/71df29ebc6872780599b7b923603c870

https://gyazo.com/7e89b468f12767daeb127e37f20d520b

https://gyazo.com/371464a4b0021d686970046044885d15

[emosun]

well i'll admit the last time i used an r9 290 was a few years ago let me check on a different machine how it's done

[emosun]

hmm I guess they changed it in later drivers as overdrive is still on a different machine that I have

perhaps you can downclock the gpu in that wattage manager program  they have on there.

[noteffy]

15 minutes ago, emosun said:

hmm I guess they changed it in later drivers as overdrive is still on a different machine that I have

perhaps you can downclock the gpu in that wattage manager program  they have on there.

Global WattMan tab? Thats for graphs showing activity...temps etc, cant change anything besides deselecting  not to show certain graphs.

[emosun]

maybe you have to download overdrive now heres the page https://www.amd.com/en/technologies/amd-overdrive

[noteffy]

https://gyazo.com/dbfa138dc95e0864f73496e42da3debc

 

[noteffy]

13 minutes ago, emosun said:

maybe you have to download overdrive now heres the page https://www.amd.com/en/technologies/amd-overdrive

https://gyazo.com/dbfa138dc95e0864f73496e42da3debc

 

[noteffy]

Have since replaced my PSU and the problems have disappeared.

 

P.S I also replaced my motherboard, cpu and ram.

 

Have not changed GPU. From all the evidence, it was the power supply.