R9 290 or the PSU reboots the PC

Knut1337

Commendable
Sep 28, 2016
10
0
1,510
Hello my friends,

before you ask: Yes, I am aware that there are plenty of threads featuring the same topic. Unfortunatly I spent now about 6 days of reading them here and on other sites.

I currently suffer random reboots of my system. It started about a 10 days ago while playing cities skylines. In the meantime it also crashes while playing an even older game (cities in motion). However higher demanding games like ArmA 3 & GTA Online are running quite well - except the latter one crashed 2 times in about 80hrs of playing.
Cities Skylines crashes within 10 minutes now.

In the beginning I was not sure where the problem is, so I tested a lot of stuff & finally opened another thread which should be closed already. However the long version can be found here:

Hello, I am not sure if this "systems" is right , but I thought as it is about the whole system it should be okay.

As one might guess from the titel, my pc crashes after playing between 10 & 20 minutes cities skyline. Sometimes also randomly (and after more than 2-4hrs) while gta online, but that only happend 2 times. But it still happend.

So there must be anything wrong with my system that little voice in my head told me and I startet to do all the regular stuff.

I updated my drivers - graphics, chipset & audio.

I ran memtest for about 2-3h and it didn't reveal any errors

I ran prime95 for about 7h and it also didn't reveal any errors.

I also bought a new psu about a year ago so I am pretty sure it is not about that, even though I back-checked it with the "psu calculator" of coolermasters - not sure how much that is worth, but after the results it gave me, it shouldn't be the problem.

My current specs - no overclocking going on, I am such a decend guy..

AMD Radeon R9 290
4096 MB / GDDR5
Windows 10 (64 bit)
12 GB RAM DDR3 - 2x 4GB; 2x 2GB
Intel(R) Xeon(R) CPU E3-1231 v3 @ 3.40GHz
ASRock H97 Pro4 Mainboard
Coolermasters G750M

So yes, 2 different RAMs. The 2GBs are from Kingston the 4GBs from Crucial Technology.
To be "sure" I switched them between the slots and used them together and single (meaning CT& Kingston together or only CT/ only Kingston). Made no difference.

I am pretty much running out of ideas of how to find out where the problem is. Any suggestions out there?

Short Version:


AMD Radeon R9 290 (Club 3D)
4096 MB / GDDR5
Windows 10 (64 bit)
12 GB RAM DDR3 - 2x 4GB; 2x 2GB
Intel(R) Xeon(R) CPU E3-1231 v3 @ 3.40GHz
ASRock H97 Pro4 Mainboard
Coolermasters G750M

Bought everything in spring 2015.

I tested the RAM, CPU (prime95) and nothing brought up any error.
I updated the drivers, which did not helped. I tried it with older drivers, which did not helped either.

Then I started to cap my fps, after that I found a thread suggesting to force constant voltage and finally found one suggesting to lower the Core & Memory Clock a little bit. Nothing of that stuff helped.

Finally I was able to borrow an older GPU yesterday (GTX 550Ti) and ran Cities quite smoothly for about 2hrs.
That led me to the suggestion that either there is something wrong with my GPU or my PSU is simply not giving enough power - which cannot be excluded as the GTX is not demanding as much as the R9 290.
However 750W/12V should be sufficient I think?

So I ran the 3DMark DirectX11 Benchmark today and logged my GPU state with GPU-Z.

My system rebooted in the third test I think? (the one in the jungle).
However the log shows that it crashed after the high demand/ stress scene, at least that is what I read out of it. I attached it anyhow.

The test starts about 13:47h.
That suggests that it is not the PSU right?

I just want to be sure as the higher power PSU is about as expensive as a comparable new card/ the same again. So it would be cool to know that it is surely the GPU and I don't have to buy both in the end.

Glad for any help.
 

spat55

Distinguished
Could be PSU and I always keep a cheap but reliable one for spare to test with. It could also be unstable GPU and I'd download MSI Afterburner and underclock it just to find out, if you run it at a really low Core/Memory speeds and it doesn't crash that would be a slightly better indicator. Try wiping drivers with Guru3D DDU although I bet it's the PSU.
 

spat55

Distinguished


I'd say for a R9 290 any decent quality 550w would do the job but I'd want a 650w minimum if it was me.
 

Knut1337

Commendable
Sep 28, 2016
10
0
1,510
Could be PSU and I always keep a cheap but reliable one for spare to test with. It could also be unstable GPU and I'd download MSI Afterburner and underclock it just to find out, if you run it at a really low Core/Memory speeds and it doesn't crash that would be a slightly better indicator. Try wiping drivers with Guru3D DDU although I bet it's the PSU.

So I clocked it down again and ran the Benchmark. Now it came through the whole test. That leads to the PSU?
I clocked down Memory and Core 100MHz just to test the Benchmark again - even though that did not help with Cities Skyline.
 
maybe use hwinfo64 that shows min average and max of a lot of things like psu power and see after you run testa and see issues how them results stack up you may see a big dip like in psu power at say the 12v+ that falls below ATX spec ?? or anything else it could show you

downclocking the card if box stock is not great any card should run fully at factory stated specs issue free

http://imgur.com/N4V6t5Q

http://imgur.com/AvRnRgU

don't know if a AMD card shows this ?? [way too much orange in that field ]

http://imgur.com/YDTq3pV
 

Knut1337

Commendable
Sep 28, 2016
10
0
1,510


Thanks for the tip with HWInfo. I logged it & started the Benchmark with the factory clocks. It crashed again.

I have a GPU Power and AUX Power, don't know which one is the one that is shown at your screenshot. However both go up and at the same time the +12V as well as the "GPU VRAM Voltage In" goes down a little, as seen on this screenshot.

The Log as .csv

I could not find the last sensor of yours in my GPU-Z, so I guess AMDs don't show it.
 

Knut1337

Commendable
Sep 28, 2016
10
0
1,510


well the +12V did not drop below 11.88V so that is within the tolerance range.

The GPU Core Voltage ranges somewhere btw. 1.2 & 1.193.. I guess that means the PSU is all right?


 
It's impossible to know what's going on with PSU voltages. Firstly, software stinks most of the time for troubleshooting this stuff. Secondly, when you are gaming, load is constantly changing and is going all over the place. There are transients that the PSU has to respond to, that causes large voltage spikes, that can also increase ripple for a brief period of time. There is a ton of stuff that can happen that normal 100% load stress tests will not show.

Is your PSU top-mounted? And if it's bottom-mounted, is the fan facing down or up?
 

Knut1337

Commendable
Sep 28, 2016
10
0
1,510


Bottom mounted and the fan is facing down (yea that is indeed the worst combination, isn't it?). But still it occured after 1 year of running quite smoothly.. However - after the reboot - I shut it down once and "checked" the temperature at the bottom, it did not seemed to be to hot. But obviously that is not really a satisfying way to test it.

 
''It's impossible to know what's going on with PSU voltages.'' Power supply Masterx lol.....

''+12V did not drop below 11.88V s'' better then my seasonic x850 with a 980ti fire strike pull 265w to the card and the 12v+ dips to 11.69


''Bottom mounted and the fan is facing down '' fan should be up to draw air

do you have a buddy to test your card in his computer to see if ti does the same or any other issue ?? if in it and works fine may not be the card or have a buddy who got a extra working psu to try in your build to see ??

tyhing is like I say them top R9 cards were known to be /get this way and it was working fine for the first year then just went bad ?
 

Knut1337

Commendable
Sep 28, 2016
10
0
1,510


Well my case implied it to build it with the fan downwards, but now writing this and you saying it.. stupid idea.. I will turn it around

Unfortunatly the only buddy I know nearby has an even weaker psu, so that probably won't help.

Yea it just started to reboot about 10 days ago. Before it everything was fine. I bought all the parts in february 2015.
 

Knut1337

Commendable
Sep 28, 2016
10
0
1,510


okay, I guess for now I leave it as it be. However I tried another round of Cities in Motion and it still crashed even though the Voltage did not drop in a strange way - I logged the session with HWInfo again.
The +12V range btw. 12.1 & 11.98V. Average: 12V; Value at Crash: 11,99V
The GPU Voltage In ranged btw. 11,88 & 11,75; Avg. 11,81 Value at Crash: 11,78V
GPU Core Power: 5W-92W; Avg. 19,91W VaC 32,99
GPU Power Out: 46W Max to 4,4W Min (Idle) Avg. 20,6; VaC: 19,86W
GPU Power In is pretty much the same.
GPU VRM Voltage 11,75V - 11,88V; Avg. 11,81 VaC: 11,78V

For my foolish eyes that looks like pretty solid stuff coming out of that PSU?
VRM Temperatures at Crash were at Max ~50°C; GPU 55°C

 

Knut1337

Commendable
Sep 28, 2016
10
0
1,510


so that leaves me pretty much at gambling to buy either a new psu or a new gpu and hope that is the issue?

would the EVGA 220-G2-0850-X2 fit then?
 
Sort of a gamble I guess. Chances are, though, that if the software voltages are that close to 12V, then they probably really are. I'm just saying it's not really a way to guarantee that nothing is wrong, things can still happen behind the scenes. I don't really know if it's your GPU or PSU at fault.
 
ya, its a tiugh call on what to do ? maybe psu ? maybe card ? maybe something else in the motherboard like a unstable or bad vrm ??

that why if you don't got any way to test things like a back up card - psu or computer all you can do is guess if nothing up front point to something ?

you know it was working fine but now failing ?

''However higher demanding games like ArmA 3 & GTA Online are running quite well - except the latter one crashed 2 times in about 80hrs of playing.
Cities Skylines crashes within 10 minutes now.''

see this could be a program or software issue as well or a hard drive failing / corrupted data / ????

I had asus suite on a asus board that did all this I never knew what kind of computer I had one min to the next in till I figured that out may go 1 min may go all day may go 4 hrs ?? amd all acted like hardware / memory issues I was having fits - swapped cards psu's memory even cpu and ended up that asus program ..


I got to admit my nice expensive at the time seasonic x850 psu dropping like I showed did not impress me at all

I don't know on this ?

GPU Power Out: 46W Max to 4,4W Min (Idle) Avg. 20,6; VaC: 19,86W
GPU Power In is pretty much the same.

at the screen shot I gave with gpu power highlighted you see my 980 ti went up to 256w at that full load use ? seems to me that R9 290 would be close if not more [running firestrike ]

have you tried it and monitor the readings to see how it pulls things down ?? its a bit more demanding then a game

http://www.futuremark.com/benchmarks/3dmark


''The fan should face downward as long as the case has holes there for it to suck air in from. If it faces up then it sucks in hot air''

todays psu's should not be getting that hot an should run near -0- fan and then if the fan is running it will draw hot ait out of the case and expel it out like a case fan would