Help please! Troubleshooting reboot. PSU/MB/Ram/GPU

Uma Debray

Distinguished
Jul 18, 2015
12
0
18,510
Hi I've been reading through threads here for a couple of days trying to resolve this issue, I haven't *found* it already answered, but I'm not very experienced so apologies if this is something obvious that i've just not searched for properly!

System
I think this contains all the details of my system on this CPU-Z validator https://valid.x86.fr/g5ek4l

my PSU is a :XFX - XTR 650W 80+ Gold Certified Fully-Modular ATX Power Supply

case is a ENTHOO PRO TEMPERED GLASS

CPU: Intel® Core™ i5-4690 CPU @ 3.50GHz

Cooler: Cooler Master - Hyper 212 EVO 82.9 CFM Sleeve Bearing CPU Cooler

GPU: R9 380

MB: Gigabyte - GA-H97M-D3H

Ram: Corsair - Vengeance LP 16GB (2 x 8GB) DDR3-1600 Memory

boot drive: SanDisk - SSD PLUS 480GB 2.5" Solid State Drive


Fault
I've been getting noticeable drop in frame rate in some games that when I'd previously played them had run very well (bf1, Overwatch). I tried playing Destiny 2 for the first time in a while, and while it ran beautifully before, now a few seconds after leaving the menu and entering the game space the whole computer reboots (I looked in event viewer and it just registers an unexpected shutdown).

what ive tried
I've tried this a few times with MSI afterburner running so i could monitor temp (as previously this card was overheating until i replaced the thermal paste) and temp didn't pass 80c and often cut off far lower.

I've had trouble with ram before, so I tried different combinations of the two sticks in different slots. the PC wont get as far as the bios (is this what is meant by "post"?) without them being in the slots listed on the Validator image (weirdly, these are labeled "2" and "4" on the board/manual and are adjacent slots right next to the fan). I can boot with one stick, and get the same results in game test.

I ran memtest overnight and no faults with the ram sticks

I've tried monitoring voltage and at idle i get these results:

vcore: 0.8
12=12.24
5=5.07
3.3=3.344
vaxg=0.012
vrin=1.76

I ran prime 95 and voltage didn't move much at all and no failures

I ran furmark and voltages did move but no were near the thresholds advised on another thread and temp stayed in the mid 80s. after five minutes I closed the windowed furmark and on doing so the computer dropped power and rebooted.

In the validator image i noticed the frequency was highlighted. I haven't ever overclocked anything, so i was concerned about this. I installed intel extreme tuning and disabled turbo boost.

ran furmark again and same happened.

at this point I disabled the second monitor (another 21") thinking maybe this was drawing too much power (i havent always used it) and on clicking the change to single display in windows the computer dropped power again and rebooted.

Finally, i've just taken the GPU out and run furmark through the onboard graphics with no trouble.

Thanks for reading all this! I've probably done some of this trouble shooting in the wrong order! Hope someone can help!





Update: I was really hoping this was going to be a fix I could get to without putting hundreds of pounds down, if only temporarily, as individual tests on PSU, GPU and memory are all coming back good. I'm not sure how else to check the MB, its definitely playing up with regards ram placement. I'm looking at ordering a new psu and a new gpu. part picker thinks these will fit:

https://www.scan.co.uk/products/evga-geforce-gtx-1060-sc-gaming-6gb-gddr5-vr-ready-graphics-card-1280-core-1607mhz-gpu-1835mhz-boost

and

https://www.ebuyer.com/630112-evga-supernova-750w-fully-modular-80-gold-power-supply-220-g2-0750-x3#

 

Uma Debray

Distinguished
Jul 18, 2015
12
0
18,510


Thanks, that would seem like a good way to do it, unfortunately I don't have access to either of those so it would be a matter of buying one
 

Yes - you can make use of the return policy if necessary.

 

Uma Debray

Distinguished
Jul 18, 2015
12
0
18,510


I'm afraid I don't understand. Are you saying ive installed the ram wrong, or that its the wrong ram, or something else..?

edit: I've not heard of this. seems my mb just needs the ram in slots 1 and 2 for it to configure as dual channel. unfortunately the computer didnt boot last time i had them in this combination
 

Uma Debray

Distinguished
Jul 18, 2015
12
0
18,510


hmm, ill have a look and see if thats an option.ill have a look and see if there's a good upgrade for the 380 that i can afford so if it does solve the issue i could potentially hang onto it. I'm wary about returns
 

I'd start with the power supply. Any store will allow you to return it.

 

Uma Debray

Distinguished
Jul 18, 2015
12
0
18,510

thanks for your continued help!
it seems i had a old gpu i forgot about!
its an hd 6670
I just tried the furmark with prime95 at the same time for 10 mins with no fail.
it doesn't draw power direct from the psu by cable though, so does it not failing say anything about whether the 3 year old psu does or doesnt work properly?
 

Unfortunately that isn't a valid test, as the 380 draws a ton of power. You don't even have to uninstall the preexisting power supply. Just turn the case over on it's side, unplug all of the PSU cables and plug in the temp replacement psu into the ports on the board and any internal components. That way if your system still crashes and it isn't the power supply, all you have to do is reconnect the cables. Just make sure the PSU doesn't fall into the case.

 

Uma Debray

Distinguished
Jul 18, 2015
12
0
18,510


I'm actually not comfortable buying something to test and then returning it, as this is specifically against all the t&c of the shops ive seen so I could really easily end up with a £100 of psu i dont need depending on how they respond.

I had been getting normal voltages during the furmark test, and reboot occurs only when i tried to close that test. Is there any other way to test if this psu has a fault ? I thought maybe HWiNFO would be able to record voltage at the point when reboot occurs and thought this could be useful, but I can see how to make that happen.
 

All you have to do is find out if they have a return policy. Unfortunately the only way to effectively test a power supply is by temporarily replacing it.
 

Uma Debray

Distinguished
Jul 18, 2015
12
0
18,510


Yeah, I've checked the returns policy of places I could buy from in store, pc world (unopened packaging), local computer shops have either no returns without fault or discretionary £10 testing charge and still no guarantee they would take it back. Standard distance selling laws allow 14 days cooling off period and I could even get around the cost of returned postage by buying online form pc world and returning to store in person but those t&c are quite specific regards "unused" because its meant to account only for the fact that in store you could ask for a box to be opened. You might no of somewhere else that has a more lenient policy but ive not found it. Amazon for examples is the same, but they extend the duration to 30 days.

Again thanks for your time and help, but leave this thread open for anyone else that might have another suggestion about diagnosing a fault with a psu other than replacing it.
 

10-4 Good luck.