PC shuts down or reboots spontaneously (MSI Eclipse SLI)

Question

Starting a new thread to avoid totally hijacking another thread talking about almost the same symptoms here.
My original post:
Quote from: Smoke on 05-April-09, 12:43:21
I get this same situation.. except the event viewer doesnt state anything.
The system is not bluescreening.. it just hardware powers down and reboots.
I get the same message as the original poster as well, it is very annoying as it often happens in the middle of doing something.
I wonder if it could be thermal related to the high IOH temps on the board (often 80-85C, but not sure what happens to them in the middle of playing a game)? Is there any way to log the temperatures to a file so I can check what they were at the time of the crash?
I get this issue whether I change the default bios settings or not. Ive got a 920 on my eclipse board, my CPU temps are below 30C under load. System temp is about 27C.
What can I do?
The following replies:
Quote from: AaronYuri on 05-April-09, 12:56:57
That's way too high, even for 100% load.  Reinstall the heatsink.
Quote from: Smoke on 05-April-09, 20:46:53
It is? I found another thread on this forum where it was mentioned that intel and msi said the IOH runs hot and to expect temps up to 100C. How do I go about changing the manufacturers heatsink on a board? Ive never done that before and cant find any procedural help thru google...
Ive been able to replicate the issue by running ram up to high usage. When I get above 98% ram allocation, I spontaneously reboot. So It could be a ram issue as well. Of would that be something possible IOH related as well?
Quote from: NovJoe on 05-April-09, 23:37:33
smoke,
Can you list out your full system spec and PSU spec?
System Spec:
Core i7 920 w/ TRUE 120
MSI eclipse SLI (bios 1.4)
6x 1gb corsair 1866mhz rated ram sticks (TR3X3G1866C9DF)
1500w thermaltake psu
galaxy gtx295 in the primary x16 pci-e slot
MSI eclipse soundcard in the 1st pci-e x1 slot
4x 1tb samsung 'UJ' hdd (raid 5)
O/S (= old winxp 32 bit until I get hold of a 64 bit os) running on 128gb PQI SSD
lg bluray burner
All the drives are sata2 and running on the ICHIOR ports.
For bios settings Ive got the optimised defaults with spread spectrum off and turbo boost on. Once its table then I will OC, but for now its not stable at the baseline yet.
Temp info:
IOH temps are as I said, around 80 at startup but rise to at least 87 under load, but not really much further that Ive seen.
CPU temps are very low with the dual-fan TRUE 120 heatsink. I havet seen it go above 35C with a reasonably heavy processing load. Usually sits about 0-2C above ambient at most when idle.
System temps get to about 40 under load, or around 27-35 under regular usage, but Im not quite sure what thats measuring the temp of on the board.
The issue is very frustrating but somewhat intermittent - I can go two days without a crash but then get several in a row. I wasnt previously paying attention to my usage at the time tho so I dont remember what I was doing when it crashed that I wasnt doing when it was fine. I have been able to replicate the crash/reboot/shutdown symptom by running multiple memory intensive apps and loading the ram up to just on the 2.99gb mark (the most winxp 32 bit will 'see') and it goes down. No blue screen though.. its a hardware issue not software I am pretty sure.
[edit] Updated the ram part number to be correct.

Answer 1

ok, should I label this possibly as solved?
I followed MSI's advice and checked through on drivers. Ive got totally up to date drivers, but nvidia just put out a .5  driver update (i mean, its increased the driver version by .50) and so I updated to that. I still got a sft reboot on testing, so no real joy.
THEN, I did the big change MSI asked me to do - Ive successfully brought my system up to bios 1.5B3 and pow, massive improvements all round!
Intelburntest runs perfectly now, no crash, no reboot, no errors of any kind. All residuals match. No memory errors. Memtest86 2.11 runs smooth as silk without errors either. PCmark runs out a basic benchmark of almost 19000 (about 3000, or 18.75%, more than the std core i7 920 benchmark without OC'ing) and looks great. Couldnt get that to run previously either.
The previous methods which 100% of the time would replicate the soft reboot of the system now no longer do so at all (i.e. 0% of the time yet) and for that matter I have not been able to get the box to fall over AT ALL since the new bios.
I have the same temps on cpu/ioh/system as I did previously. I recall for the same clock speed the cpu was putting out something well under 20 gigaflops... now its a solid 28. I didnt even realise that was a symptom of anything wrong, else the bios does some amazing extras.
I should note, Im only now using the 'optimised defaults' setting thru the bios. No overclock at all yet.
I'd love to know what MSI changed from 1.4 to 1.5B3, but whatever it was.. it does appear to have solved my issues. I wont say that conclusively just yet, but I cant bring the system to a crash no matter what I try now where it was rather easy to blow it over with a quiet whistle before.
Other things Ive noted is that the page file now seems to be actively utilised. Ive pushed it at load up to almost 3gb where before the bios update I was struggling to see 400mb used. Not sure if that was part of it, but if you get simiar issues go straight to MSI and see about getting hold of the 1.5B3 bios or later.
To quickly address the recent replies:
1) Ive got to agree with HU16E with my ram. I have no trouble running at the 1333 JDEC spec. And I feel very positive about now being able to achieve the XMP spec with this new bios.
2) Thanks for the link Aaron. I did end up using the USB flash method from this forum. Worked perfectly. Chalk up another success story!
3) I did try to swap out the ssd last night, but the system kept soft rebooting while installing the OS on the spin drive. SO I gave up. Seems the ssd isnt a drawback under the new bios tho (assuming it even was a cause earlier).
4) Ive DL'ed the x64 windows 7 beta and I'll likely as not have a shot installing that and refitting the extra 3gb ram back into the system. I'll let you know how it goes. Theoretically 6x1gb sticks should be as much as 15-20% faster than 3x 2gb sticks (and were about that much cheaper for me at time of purchase) but I'll run some benchmarks now that the benchmarks work and let you know the differences in case anyone wants to build their box the same way.
Conclusion:
I'll stress the system over a bit for the next few days without overclocking and I'll see if I can bring the system to reboot like it did previously through high usage, and if I cant I'll change the thread to solved and write out a proper solution post for the problem, symptoms and solution.
Id like to take the opportunity to say thanks to all the guys who've given me help and advice. Youve all been really great. Thanks a bundle! I owe you one