Sign in with
Sign up | Sign in
Your question

New 3930K + 48GB system BSODs

Tags:
  • CPUs
Last response: in CPUs
Share
December 11, 2011 6:34:32 PM

OK, so I bought this system as a poor man's server:
3930K
Asus P9x79 Pro
3x G.Skill Ripjaws-X 2x8GB (=48GB)
Asus GT210 512MB
Antec High Current Gamer 750W
Intel 80GB SSD
G.Skill 40GB SSD
WD 500GB 7200rpm

I get BSODs quite frequently, about twice a day. Funny thing is, I'm not able to reproduce using any torture test I know of. I tried Intel Burn in, Prime 95 on each setting - ran for hours and i also ran some tests from Sysoft Sandra. For each of these tests, I also copied files from drive to drive - no BSOD on any test.
I tried to remove 2 sticks of RAM (down to 32GB) and I still get BSODs in about the same frequency. The machine is running SQL Server on Windows 2008 R2. The crashes usually happen when I'm running an intensive query or when restarting the SQL Server process.

I'm looking for a test which I can run that reproduces the crash so I can start to investigate. Also, if anyone has an idea what can cause this - I'm dying to solve this. I suspect the PSU, the board memory compatibility, or maybe it's a faulty mem stick, I'll keep switching them to find out. The rest - I don't have spares lying around (who does). The memory doesn't appear on the motherboard mem compatibility list - what does that mean?

Thanks for anyone's help!

More about : 3930k 48gb system bsods

a c 900 à CPUs
December 11, 2011 6:38:03 PM

All you need is to have one bad memory module in there to get BSOD's.
m
0
l
Related resources
a c 92 à CPUs
December 11, 2011 6:45:06 PM

use memtest86+ and run multiple passes. run the manufacturers hard drive diagnostics for all your drives. also, what is the blue screen error??? that would give you a clue. if i suspect any of the components it would be ram or a bad hard drive. the fact that the ram is not on the motherboards compatable list is a very likely problem. ram errors may not even show in memtest but it could be the problem.
m
0
l
December 17, 2011 11:33:52 AM

I ran memtest86 for several hours without finding any error. I can't find the BSOD info in the Event Viewer but I have set the system to write a minidump next BSOD so I will look at that. (The server isn't used by anyone locally, only remotely)
m
0
l
a c 152 à CPUs
December 17, 2011 12:38:28 PM

Did you turn off Hyperthreading?

SQL server doesnt like HT fwiw.
m
0
l
March 18, 2012 4:30:13 PM

A few months went by, BSODs went away and came back. I ran memtest86 again, got errors, didn't stop test and after a while this happened: (see at end)

I'm now going to test each pair of sticks individually, hopefully, there is one bad pear :) .
If not, is there a voltage setting I can increase which might improve stability?

m
0
l
March 18, 2012 4:45:16 PM

Hi, I had a look @ G-Skill SSD. Does your G-skill SSD classified as Pheonix Pro? If so, that has SF-1200 chipset. Ouch...

But hey, it may be memory issue other user stated and you found some errors. :) 

m
0
l
a c 127 à CPUs
March 18, 2012 5:44:31 PM

ytoledano said:
A few months went by, BSODs went away and came back. I ran memtest86 again, got errors, didn't stop test and after a while this happened: (see at end)

I'm now going to test each pair of sticks individually, hopefully, there is one bad pear :) .
If not, is there a voltage setting I can increase which might improve stability?

http://s17.postimage.org/9jy6bf2fj/memtest86_error.jpg


Thats a old version of Memtest. You should be using 4.2, in fact it wasn't till v 4.2 that we could get it to work on SB based CPUs properly so I am suprised v 4.0 works.

The fastest way to find the bad memory stick is to pull them all out and test the each individually. Of course it may take some time as the way Memtest works is it randomizes some tests every pass so it may not hit a fail until pass 3 or later.
m
0
l
March 18, 2012 6:20:48 PM

You're talking about memtest86+ which is at version 4.2 (I used memtest86 v4.0). I will use memtest86+ for my further testing.
m
0
l
!