Mysterious reset with no warning or BSOD

[ForceGaia]

For the past few days i've been encountering problems with the stability of my computer. I've no idea what has caused it, so my theory is the something inside is about to fry. I can be working normally and the computer will reset with no warning, no BSOD (bar one case) and no logs bar a standard kernel system-power that you get on any hard shut down.

I've done some tests already, and found some results but could use some help interpreting them:

My specs - nothing overclocked unless specified:

• 
  ■ Win 8.1 Update 1 Pro
  ■ AMD FX-8350 8-Core processor @ 4.0Ghz
  ■ 4x 2GB Corsair XMS3 1333Mhz DDR3 ram
  ■ Asus M5A97 R2.0 Motherboard
  ■ Gigabyte 2GB Nivida Geforce GTX 660Ti Graphics (factory OC'ed)
  ■ 5 Hard disks of varying ages and vendors (320GB, 500GB, 1TB, 2TB, 3TB)
  ■ Creative Sound Blaster X-Fi Xtreme Audio PCIe
  ■ Corsair TX 650W PSU

Tests performed and errors received:

• 
  ■ Chkdsk with /offlinescanandfix flag on all drives. all bar 1TB drive came back clean. The 1TB drive had some filesystem errors which were fixed, no bad clusters.
  ■ Windows memory Diagnostic - standard test - No errors.
  ■ Prime95 Torture Test - Preset 'Blend' Test - This gets nowhere and a reset occurs almost immediately.
  ■ Memtest86+ bootable CD - got to test 8 on the first pass and seemed to get stuck, wouldn't respond to any keypresses.
  ■ One instance of a BSOD when using chrome (2 tabs - Facebook and Kitguru) - error code IRQL_NOT_LESS_OR_EQUAL
  ■ 3DMark Demo on steam - Sky Diver - "An Unknown Error occurred: Unexpected error running tests. Workload Single init returned error message: Process exited unexpectedly"
  ■ 3DMark Demo on steam - Sky Diver attempt 2 - Caused reset immediately.
  ■ memtest86+ attempt 2 - not finished yet but it's 80% through and it's found 51 errors

I've asked for my multimeter to be sent to me so i can check my PSU.

I think my RAM is the culprit, could anyone else pose some potential causes?

 

[Beezy]

The fact that you leave out the power supply is a bad sign, but this is usually a power supply issue. Brand and model of the unit please?

 

[ForceGaia]

ah, simply forgot it, edited it's a corsair tx 650w PSU

 

[Beezy]

ah, simply forgot it, edited it's a corsair tx 650w PSU

Let us know when the meter arrives, RAM or PSU is very likely. but RAM generally gets a BSOD

 

[ForceGaia]

I'm running memtest86+ again as i was going out and thought i might as well see f it changes anything, and it's currently at 80% through the first pass but has found 51 errors. Would the PSU cause this effect too?

 

[ForceGaia]

i did think that a memory fault would give a BSOD as I bought some new memory and it kept doing that and it failed even the windows memory test. so i RMA'ed it

 

[Beezy]

If youve gone through two sets of RAM sticks and they both had a lot of errors then the motherboard or CPU could be at fault (the memory controller on the CPU could be busted, but thats kinda rare.) but the motherboard could be the cause? can you try the RAM in a different computer and re run memtest 86

 

[ForceGaia]

No that was an actual problem with the new ram

I was adding more by replacing two modules and immeadiately got tons of BSODs, removed them and went back to my old modules and everything was fine

When I RMAed them the company did some tests and found them faulty as well.

 

[ForceGaia]

So interesting development:

I've just been testing individual slots and pairs of RAM by basically putting each pair in each pair of slots on the motherboard one pair at a time. I was hoping this would reveal a culprit memory pair or motherboard slot. In each instance i ran prime95 to see if it would reboot, it did not.

I've not put all four modules back in, and ran prime95. It's running fine. It sounds like one of the memory modules wasn't seated correctly, although why that would become apparent now i have no idea.

I'm about to try 3dMark as btoh that and prime 95 were dead certs to force a reset.

I'll still test the PSU when the multimeter arrives on monday.

 

[Beezy]

So interesting development:

I've just been testing individual slots and pairs of RAM by basically putting each pair in each pair of slots on the motherboard one pair at a time. I was hoping this would reveal a culprit memory pair or motherboard slot. In each instance i ran prime95 to see if it would reboot, it did not.

I've not put all four modules back in, and ran prime95. It's running fine. It sounds like one of the memory modules wasn't seated correctly, although why that would become apparent now i have no idea.

I'm about to try 3dMark as btoh that and prime 95 were dead certs to force a reset.

I'll still test the PSU when the multimeter arrives on monday.

another thing to consider is the memory controller on the AMD FX chips are sub par compared to intels memory controllers. Running 4 sticks usually means having to run them at a lower Mhz because it adds too much stress to the memory controller. So 2 sticks at a higher Mhz and CAS with less stress on the memory controller might be a better thing. i can only imagine needing to use all 4 slots if using 32 Gb RAM for super high end 3D renders or something of the like.

Then again theres the 'upgrade' solution where you HAVE to use all 4 slots just to have 8 GB which is the boat youre in. Youre options are either upgrade to a 2 x 4GB kit for 60-70$ OR try setting the the RAM in the BIOS manually to have;

A. Looser timings, leave them at auto.
B. Higher the Dram voltage manually to 1.495V or possibly 1.5v (but not higher) to ensure stability