CPU shutting down PC abruptly when loaded and OC'd

I had another topic running before about this problem, but that's when I thought my old PSU was the culprit, so I bought a new one thinking that would solve the problem, but it didn't

Intel i5 2500K @ 4.6ghz in offset mode (1.35V max in offset load, same voltage as it would require in fixed mode)
Asrock P67-Pro3 Rev. B3
12GB ram divided between 4 sticks @ 1333mhz
Sapphire Radeon HD7950 @ 1155/1500 with 1193mV on the core
Corsair RM750x (previously Enermax Liberty 500W)
1TB internal drive, 2TB external drive with own PSU + 1TB external drive which gets power from the USB 3.0 port
5x120mm fans + 1x80mm fan

So what's going on? A week ago my PC started shutting off during games like BF4. Not rebooting, no, this was a hard shutoff, as if the power had been pulled. No BSOD, no error dump, literally nothing.

I've done some tests to try to pinpoint the issue and this is what I have so far

- When only running the Unigine Valley benchmark the PC doesn't turn off, doesn't matter if the GPU is overclocked or not
- When running prime95 the PC shuts off somewhere during the 1st heat up phase, but only when it's overclocked!
- Increasing the voltage for the OC doesn't matter at all. Changing voltage from 1.35V to 1.4V does nothing. On 4.2Ghz using 1.35V same issue
- Without an OC this doesn't happen!
- Max temp gets to 72c whilst average is 69c, it has been like this since forever.
- Tested memory with Memtest86 for 13 HOURS without any errors.

So right now I'm a bit at a loss here. Is it CPU, Mobo, quite possibly even the RAM (wouldn't be the first time something passed on Memtest but failing in normal operation)???
This behaviour is something I haven't come across before, so I have no idea what to do.

The symptoms do not point to ram, however, the mixing and matching of capacity can cause instability. Try increase ram voltage and/or VccSA/VccIO and see if that helps.

Have you updated to the latest BIOS? What other testing was done?

Here's the initial thread

http://www.overclockers.com/forums/showthread.php/770994-Is-my-PSU-failing?p=7918573#post7918573

Half tempted to combine them since its the same problem.....

Yeah but wasn't sure it would get attention in PSU forum
I'd still pull the 2x2 GB sticks and give it a try. Nothing to lose right

EarthDog said:

The symptoms do not point to ram, however, the mixing and matching of capacity can cause instability. Try increase ram voltage and/or VccSA/VccIO and see if that helps.

Have you updated to the latest BIOS? What other testing was done?

Click to expand...

I forgot to mention that I have upped the RAM voltage to 1.54V on stock clocks, loosened the timings a bit as well to see if that's the culprit (this ram doesn't like to run on SMP timings and has to manually be set a bit looser because of the mixing).
BIOS is the latest version
No other testing than this. I have unplugged all unnecessary power sources to eliminate the chances of a short overloading the PSU, but I figured if that was happening I would've smelled some burning wires (been there before ).

System has been running stable now without OC for a couple of hours. But then again, the system runs fine when I'm not loading the OC.

Now the only testing I can think of is pulling all the sticks, trying each slot individually and then trying each stick individually. That's going to take a lot of time though

Johan45 said:

Yeah but wasn't sure it would get attention in PSU forum
I'd still pull the 2x2 GB sticks and give it a try. Nothing to lose right

Click to expand...

Yeah I wasn't too sure about crossposting but since I've eliminated the possibility of the PSU being the culprit I thought it would get a bit more exposure here, sorry about that .

I'd still eliminate the mixed ram right from the get go. It can be a real can of worms at times.

Ok, so I did some testing with the RAM (man, that took A LONG TIME!) and have found some odd behavior.

- 2 sticks at first caused my windows to BSOD after logging in, both sets did this
- 1 stick of 2GB would seize up the PC when trying to run Prime95. I could move my mouse but everything else was frozen (CPU-Z, Realtemp, Prime95, couldn't even Ctrl+alt+del). This happened with both of the 2GB sticks.
- 1 stick of 4GB would work, I could boot into windows, run Prime95 a bit longer than usual but it would still shut off. Here's where it gets a bit strange: Load temps on my CPU were 8~10c LOWER than usual, yeah, that's a big difference!
- Then I tried putting the 2nd 4GB stick in the 3rd slot so that it would run 8GB dual channel (1st slot + 3rd slot), but when POSTing I noticed that my mobo was only reporting 4GB instead of 8GB
- Moving the sticks over to the 2nd and 4th slot made it recognize the full 8GB

Motherboard issues perhaps?

A lot of boards prefer 2 and 4 over 1 and 3. Try you 8 gb in 2 and 4 and see how it goes. Could even be the IMC in the CPU

Yeah I only have 2 sticks in at the moment in channel 2 and 4, since that's the only way to get it to recognize 8GB dual channel instead of 4GB :S

Did some more digging around, came across a tool made by Intel (Intel processor diagnostic tool) which is able to test all the aspects of the CPU, including the IMC to see if there's something wrong internally and my chip passed with flying colors. However, the program drops my OC clocks back to stock 3.3Ghz clocks so I can't really test if it's the same when OC'd, but from reading around if the IMC is ****ed due to an OC it will show up in this program.

I still have intermittent memory related crashes when booting and logging into windows, never had it when I was running 4 sticks.

Everything is starting to point more towards my mobo, too bad those are hard to come by in 1155 sockets these days. I can't think of any other tests to perform to see whether it's my motherboard.

Have you tried re-flashing the BIOS. Could be a bit corrupt. I've messed mine up more times than I can count and it will cause all kinds of odd things to happen.

Reflashed and reconfigured, but still same issues. Was getting a lot of WHEA BSOD's on login, changed offset voltage to see if that had any effect and it did, but computer is still powering down during Prime95 no matter which test.

Usually happens around 73~76% of the first heat up phase when using Realtemp. This issue has me totally stumped.

For some reason it's not getting power if it just shuts down like that or it's an overheat/overcurrent thing. Do you have a ulps?

Have you tried checking all your connections also motherboard for shorts? Didn't you have a drive that shorted out?

Johan45 said:

For some reason it's not getting power if it just shuts down like that or it's an overheat/overcurrent thing. Do you have a ulps?

Click to expand...

Nope, checked both BIOS and power settings and both seem fine.

wingman99 said:

Have you tried checking all your connections also motherboard for shorts? Didn't you have a drive that shorted out?

Click to expand...

My hard drives weren't shorted out, their index just got corrupted. Last week I took off my cpu heatsink and discovered a ton of dust lodged near the CPU and the socket, even on the CPU and socket. Cleaned that out, tried to put everything back together just to realize my old tube of AS5 had totally dried up, so I used curry ketchup as TIM. On stock clocks, it would hit the same temps as OC'd with good TIM, but I could game for hours and nothing would happen. Definitely NOT the temps. Just replaced the TIM and temps dropped 2~4c from the last time I had it OC'd. I've checked the front of the MOBO to see if something is shorting it out but that's not the case. Tomorrow I'll take the whole thing out of the case to see if there's maybe something behind it.

It's not the PSU that's for sure now. The fan on my new PSU doesn't even spin when loading the CPU and only spins on boot, so it's not even reaching 300W to be straining the 750W PSU.

The joys of PC hardware

How do you know that the fan in the PSU only comes on at 300w?

Motherboard.

motherboard was what i'm reading also.
I have had bent socket pins case sudden shutdowns also.

Pull the CPU and carefully check the pins are all straight and clean a bit of contact cleaner and a soft small paint brush is great for cleaning them.
Also check your ram slots a bit of contact cleaner in there and use an old credit card with one layer of shop towel or something suitable wrapped around it to clean them.
A lot of ram problems are caused by CPU Pins or Ram slots.

wingman99 said:

How do you know that the fan in the PSU only comes on at 300w?

Click to expand...

From reading several reviews where they discuss this. 300~350W is needed before the fan even speeds up

Mr.Scott said:

Motherboard.

Click to expand...

Yeah, sure seems like it. These 1155 mobo's are hard to come by, especially since I have a 775 mounting bracket for my TRUE which won't fit any unless it's another Asrock board with extra 775 mounting holes. Great

- - - Updated - - -

macsbeach98 said:

Pull the CPU and carefully check the pins are all straight and clean a bit of contact cleaner and a soft small paint brush is great for cleaning them.
Also check your ram slots a bit of contact cleaner in there and use an old credit card with one layer of shop towel or something suitable wrapped around it to clean them.
A lot of ram problems are caused by CPU Pins or Ram slots.

Click to expand...

It's not the CPU socket nor the RAM from what I have tested and seen. CPU socket is perfectly straight, cleaned the contacts on the CPU when taking things apart today and they were perfectly clean to begin with. I'm going with Mrscott here on this one, but first I'll pull the board out tomorrow to test if that's the case, I think my dad has another 1155 board lying around in his PC.