View Single Post
Old 09-23-2015, 06:09 PM   #59
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 24,947
Default

Quote:
Seems like his Prime95 tests would have ruled out the RAM?
It should, and rule out most hardware but to really sign off on that it needs to run 24 hours or so. Thought it does look like Max ran it for 21 so that's pretty good and rules out a lot of stuff to be honest. Many use it for that exact reason; hardware stability testing. Typically if it can run ~n hours without a single error, it 'usually' isn't hardware that is involved with code (mobo, periperials, memory, chip etc.).

Quote:
Frankly I don't know whether it verifies data integrity.
IIRC It does to an extent because if it isn't good data, the prime calcs will result in an error. That could be missed if the bad address never gets accessed but his BSOD's are pretty darn common which I think is revealing. Meaning if he crashes that much, I'd certainly hope Prime95 would show errors after a day of testing.

The strange boot error is really the same as one of the errors in the dumps I reviewed before, so it's essentially the same thing, just occurring during boot:

C:\>err 0XC00700C1

# as an HRESULT: Severity: FAILURE (1), Facility: 0x7, Code 0xc1
# for hex 0xc1 / decimal 193 :

SPECIAL_POOL_DETECTED_MEMORY_CORRUPTION

Let's be aware that a bad memory/disk address BSOD (and specifically the ones I saw in all the dumps) weren't "I have bad data at this good address", it was specifically "I was told to access a memory address that either isn't mine or simply does not exist". The latter is akin to delivering pizza to an address, only to find it's a vacant lot when you show up; you were told to go to 123 Elm St but the good data is on 123 Maple Ave. Small but very important difference that means the address gets munged some unknown time in the past and before the crash. Could be 5ms earlier or 5 minutes earlier.

There is no way I can definitively say "it is the PSU", that's not possible but you guys seem to be ruling out everything else.

If it were me... I'd sign off on each thing tested as 'good/passed' until the only possible culprit is what still remains. One must commit to that process, otherwise, the only thing accomplished is testing the same thing over and over again hoping it's the culprit; better to eliminate one by one, never needing to revisit. If there are disk/memory tests that still need to be done, do them properly and move on. IIRC he had replaced the memory sticks early on. IOW, don't try to prove it's memory, try to prove it is not memory, rinse/repeat.

Might check the Mobo settings. Is this thing overclocked in any way? Is the CPU fan packed full of dust and seated poorly due to oxidation, yada, yada.
__________________
If it requires a null test to find it, it is by definition minuscule.

Last edited by karbomusic; 09-23-2015 at 06:28 PM.
karbomusic is online now   Reply With Quote