Ongoing: random shutdowns and instability

[Mandrewoid]

Hi LTT community!  The computer I built about 4 months ago has had issues since day 1, but things really came to a head this week so I'm going to chronicle my misadventures here along with asking some advice from those who have been here before. 

 

Starting off, The specs: 

Motherboard: MSI x99 Sli Krait edition

Ram: 32(4x8gb) crucial ballistix ddr4 2400 I think

Video: Gigabyte Radeon hd 7950 windforce 3x (bought used) 

Power supply: Corsair AX850

 

[DankLord]

I'm not sure if its the same as my problem but my computer had random shutdowns due to too much power draw, I had to upgrade my PSU. Does it say anything when you boot it up again? 

[Skiiwee29]

PSU should be more than ample. Does it give you BSOD? I would go get memtest64 and run that and check it out. Do you have an SSD? If so, check with the manufacture to ensure it has the latest Firmware.

[Mandrewoid]

Shit I just typed like 5 paragraphs and it deleted it all 

[Mandrewoid]

Because I typed this all before unfortunately now I'm going for the short simple story : ( 

 

PC would not power on when power button pressed ( on the motherboard )  - workaround for 4 months by rapidly flipping the PSU switch from 1 TO 0 and back. Then it would start fine. 

 

Noticed crashing in games, and FurMark. Resulting in immediate poweroff, no BSOD. 

 

if GPU is underclocked, no crashing. 

 

If GPU clocks are increased while under load, no crashing.

 

GPU seems stable up to at least 1100core 1300mem(as long as it is reached gradually, and while under load) 

 

after a crash this week on Wed. night, I went out and bought an EVGA 650GS  - powering on issue seemed to be solved, but stuck in boot loop. rebooting on its own without human intervention continues forever. 

 

Tried Win10 repair USB: I am very familiar with this (because of the many times it had randomly quit before, I've had to repair twice, and a previous PC of mine died in a windowsupdate) 

 

Boots to Login screen and reboots as soon as password entered. - cant get to safemode

 

Tried linux USB (debian 8, gnome, no graphics drivers) - Boot with "ultimate boot CD" run "system stability test" instant crash.

 

Tried 1440x900 DVI monitor instead of 4k DP monitor.

 

Format SSD and reinstall windows - install crashed 4 times. 

 

Removed all but 1 ram stick. Install succeeded. Able to boot to Windows. Crash during AMD driver install.

After reboot, can confirm driver installed OK. 

Another crash during x99 chipset driver install.

 

Decided probably memory problem, be it DIMM or CPU mem controller.

(note RAM not OC, XMP not enable)

 

Right now I'm just at the stage of running Memtest86 on each DIMM independently on each slot. 

Seems test 13: Hammer test is prone to causing hangups on this machine. Reseated ram 3x and now passed test once. (on this one stick)  

 

 

[Mandrewoid]

17 minutes ago, legacy99 said:

PSU should be more than ample. Does it give you BSOD? I would go get memtest64 and run that and check it out. Do you have an SSD? If so, check with the manufacture to ensure it has the latest Firmware.

Damn your signature 

[Skiiwee29]

3 minutes ago, Mandrewoid said:

Damn your signature 

Whats wrong with my Signature? I use Night theme so all text is White, thus my Signature is white. 

[Mandrewoid]

4 minutes ago, legacy99 said:

Whats wrong with my Signature? I use Night theme so all text is White, thus my Signature is white. 

LOL 

I thought it was just a blank bulleted list......

[Skiiwee29]

1 minute ago, Mandrewoid said:

LOL 

I thought it was just a blank bulleted list......

lol no no no.. I just changed it to a red color so it should be able to be seen on all color schemes. 

[Skiiwee29]

Back on topic though, if your system is hanging on memtest with just a single stick, keep trying them all. If it happens to all of them, then it sounds like the memory controller on the CPU may be going bad or the Board is bad.

[Mandrewoid]

Does anyone know how to Differentiate between a bad mem controller and the motherboard? At that point should I just take it to like Canada Computers or NCIX in Toronto and get them to test the CPU in a different board? Because I'm not that enthusiastic about having to buy another 5820K just for testing purposes. 

[Mandrewoid]

It seems that with only 1 stick of RAM in,  I can pass memtest86, but with 2-4 sticks it will crash on test13- Hammer test. It doesn't report any errors, just freezes. Also of note is that it will not run in multicore mode. Memtest will cause the computer to reboot within 10 minutes

[TheGuyNL]

I bet that your PSU is almost dead. Had about the same issues as you are talking about and replacing the PSU fixed it.

[WkdPaul]

2 hours ago, TheGuyNL said:

I bet that your PSU is almost dead. Had about the same issues as you are talking about and replacing the PSU fixed it.

OP says he has the same problem with 2 different PSU.

 

@Mandrewoid after 1 stick pass memtest, did you try the same stick in a different slot?

[TheGuyNL]

27 minutes ago, wkdpaul said:

-snip-

I missed that! Might actually be a faulty motherboard then.

Also important: did you try another wall socket? Might be drawing too much current from the wall and your PC gets low on power. If you have your monitor(s) and desktop in the same socket, try to plug them in separate ones and see if that helps. 

[WkdPaul]

5 minutes ago, TheGuyNL said:

I missed that! Might actually be a faulty motherboard then.

Also important: did you try another wall socket? Might be drawing too much current from the wall and your PC gets low on power. If you have your monitor(s) and desktop in the same socket, try to plug them in separate ones and see if that helps. 

Good one, I keep forgetting about that since I never had to deal with that!

[Mandrewoid]

I have taken 1 stick of ram that passed and tried it in each motherboard socket. I have also tried with 3 different monitors (4k, 1440p and 900p) because I originally thought it was a GPU problem. I have switched wall plugs, though I have my doubts that this could possibly help, seeing as it should be able to deliver 1500watts, and the max my system could draw is 650.

 

I have been able to boot to windows for a couple days but Ive been busy on the weekend so I haven't been able to install anything.

 

I ran memtest86 for 9 hours last night and it found 0 errors. Apparently only multicore mode crashes it. it ran just fine in single core mode

[craydawg]

I had this issue also with a corsair PSU a HX850 in my case. It just randomly started restarting/blue screening because of unstable power delivery. Barrow one from a friend or do the PSU fan jumper test to confirm.

 

Unfortunatly the AX and HX series PSU's are very hit and miss, from my readings and experience.  

[WkdPaul]

34 minutes ago, Mandrewoid said:

I have switched wall plugs, though I have my doubts that this could possibly help, seeing as it should be able to deliver 1500watts, and the max my system could draw is 650.

it's not about the watts you wiring can deliver, it's about "dirty" electricity supply. With old wiring or old wall plugs you could have unstable electricity, that wouldn't show on lamps and stuff like that but could affect a PC.

[Mandrewoid]

I see.

 

For future reference to everyone: I am now using a brand new EVGA 650GS

[WkdPaul]

18 minutes ago, Mandrewoid said:

I see.

if you want an exemple of "dirty" electricity, I think the second scrap yard wars has that, where one of the PC would BSOD a lot after they moved it, they changed wall plug and everything was back to normal.

[Mandrewoid]

after consulting stackexchange old forum threads, I have reseated my CPU, and it seems? To be running stable in windows using Passmarks BurnInTest V 8.1.  I lifted the motherboard out of the case by holding onto the CPU heatsink. Although I've seen this done tons of times, and thats how I've always done it, maybe that wasn't a good idea