• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Rig instability

Jon855

[H]ard DCOTM January 2008
2FA
Joined
Aug 25, 2005
Messages
12,132
While I don't have the exact specs within my head.

I have Dual Xeon X5550s running on a supermicro mobo - ain't in the mood to look up the exact model right now.

Memory banks are entirely populated - 12GB (6x2)

The rig will from time to time have a BSOD - dumping memory etc, etc probably on every other two days - this has been happening ever since I've had my rig. Never cared to troubleshoot it.

But now I'm frustrated.

I'll post more when I get home tonight to list the dumps and such - hopefully I'll find what's causing the issue.

With regards - I'll keep on Boincin.
 
oc? Have you stability tested it if there is one?
Have you run memtest already on the ram (oc'd or not)?

Those would be my starting points for troubleshooting. ;)
 
Since heat is ruled out, the next thing I'd trouble shoot for is memory. Run a mem test utility to determine they are OK. Also, look at Supermicro's site to find out if the modules are on their recommended list for your board model. If everything checks OK, revert the BIOS to default settings and see if the problems still occur.
 
oc? Have you stability tested it if there is one?
Have you run memtest already on the ram (oc'd or not)?

Those would be my starting points for troubleshooting. ;)

Didn't stability tested it - will run memtest on the next crash.

Since heat is ruled out, the next thing I'd trouble shoot for is memory. Run a mem test utility to determine they are OK. Also, look at Supermicro's site to find out if the modules are on their recommended list for your board model. If everything checks OK, revert the BIOS to default settings and see if the problems still occur.
Upon the recent crash - several days ago - I went into the BIOS and pressed reset to default settings and left it at that.

What power supply are you using?
Uhm... SeaSonic X750 Gold 750W

Anyways... Here's my CPUZ validation http://valid.canardpc.com/show_oc.php?id=1187183 - this shows the specs as my cpu's currently pegged out at 100%

Core Temps shows different freqs of 133x21 (2.8Ghz) whereas CPUZ shows 133x22 (2.93Ghz) and the TMonitor shows the cores pegged at 2,933.4Mhz

Uhmm...
 
Last edited:
monitor.jpg
 
In regards to the frequency discrepancy, maybe CoreTemp just doesn't recognize Turbo?

I would see if you can get a 1kW PSU from Enermax, Thermaltake, Corsair or similar and put that in instead of that Seasonic.
 
I would see if you can get a 1kW PSU from Enermax, Thermaltake, Corsair or similar and put that in instead of that Seasonic.

1kw isn't necessary for a dual cpu server board system with 1 gpu (even if the gpu is a 5870 according to his screenshot).

Secondly Seasonic makes quality stuff. A 750w by any reputable psu company (which Seasonic is) is more than enough to power that system.

I don't think it's a power issue. Let us know how memtest checks out on the ram.
 
Which Bios version are you running?
The DTL ships with 1.0 / 1.0a which is a very early revision.
Make sure you update to at least 1.0C or the newer 1.1 that was released last month.

I had some memory issues with the board as well last month.
Found out what memory dimm was fault, but would pass memtest+ with flying colors.
I noticed that Windows was showing 10GB instead of 12GB was how I found the error.

I am running the same memory in my 3x E5504 systems with the same motherboard so the memory is supported with that board.

CPUz & CoreTemp have always has the issue with detecting Turbo Mode Correctly.
CPUz is the correct source to use, as X5550 will go to 2.93Ghz (2660 mhz + 266 mhz Turbo Mode)
 
you said all the memory slots were populated right? have you tried inceasing the northbridge voltage just a tad to see if that might be the cause?
 
you said all the memory slots were populated right? have you tried inceasing the northbridge voltage just a tad to see if that might be the cause?
I have that option on my Skulltrail board but never used it. How can it help with OCing?
 
I have that option on my Skulltrail board but never used it. How can it help with OCing?

really only helps if your stability problem is memory based.. but seeing as how all his dim slots are filled its possible hes putting extra load on the IMC/NB thats causing it to crash.. if thats the case a increase in voltage typically helps.. its the same thing with the skulltrail boards.. hell even more so because the NB's on the socket 771/5's are horrible when running full dim slots.. but even so it really doesnt help that much on the skulltrail because they dont overclock worth a damn.. board runs way to hot..
 
... but even so it really doesnt help that much on the skulltrail because they dont overclock worth a damn.. board runs way to hot..
True, I get around the same or even less OC on some processors with the ST than I do with the better Supermicro boards, and Supermicros don't even have any voltage adjustment whatsoever, LOL. Skulltrail was really made for the Extreme QX9775 chips which had a lot of OC headway, more than the ST board was ever capable of delivering in practice.
 
True, I get around the same or even less OC on some processors with the ST than I do with the better Supermicro boards, and Supermicros don't even have any voltage adjustment whatsoever, LOL. Skulltrail was really made for the Extreme QX9775 chips which had a lot of OC headway, more than the ST board was ever capable of delivering in practice.


wish that was true.. no matter what my friend did with his QX9775's the best he got out of the skull trail was a 900mhz overclock.. the ST was one of those "omg its an overclockable dual socket motherboard" ploys by intel and it ended up being a bust.. which is what im wondering about with the SR-2.. the skulltrail was a beast using LN2 as well but when it came down to water cooling and air it was horrible.. so far we have yet to see any overclocks on the sr-2 besides benchmark LN2 cooling runs..
 
1kw isn't necessary for a dual cpu server board system with 1 gpu (even if the gpu is a 5870 according to his screenshot).

Secondly Seasonic makes quality stuff. A 750w by any reputable psu company (which Seasonic is) is more than enough to power that system.

I don't think it's a power issue. Let us know how memtest checks out on the ram.
It could just be he got unlucky, or it's just getting old and isn't able to supply enough power anymore. Because it sounds like a power problem. Bad memory isn't very likely to cause random BSODs, and in any case, that can be easily checked by Memtest, or simply by raising Vmem or dropping the multi.
 
i noticed your voltage is .840 seems very low to me for being under load. try 1.0 or 1.2
 
Sorry Jon, if I'm hijacking, just need to mention a couple of things... :eek:

really only helps if your stability problem is memory based.. but seeing as how all his dim slots are filled its possible hes putting extra load on the IMC/NB thats causing it to crash.. if thats the case a increase in voltage typically helps.. its the same thing with the skulltrail boards.. hell even more so because the NB's on the socket 771/5's are horrible when running full dim slots.. but even so it really doesnt help that much on the skulltrail because they dont overclock worth a damn.. board runs way to hot..
Yes, well I have a Thermalright HR-05/IFX mated to a standard aluminum sink that covers the MCH chip where the HR-05 cannot make full contact (chip larger than base of HR-05). The CPUs are cooled with a 120mm push/pull setup that directs airflow in a continuous stream: dual front case fans through dual ducts > push fan/pull fan on CPU1 > HR-05/custom hybrid HSF > push fan/pull fan on CPU0 > dual exhaust case fans.

I tried the FSB voltage adjustment, and now I'm getting slightly higher OCs with just a single increment increase so far. I could probably go even higher and assume that the limiting factor is my memory, which is just a single 4GB 667MHz Kingston module. I had better success with the 512MB and 1 GB modules so this adjustment might actually extract a bit more from my system. Would increasing the MCH voltage help any? Don't want to try something that might hurt more than help.

wish that was true.. no matter what my friend did with his QX9775's the best he got out of the skull trail was a 900mhz overclock..
I read the main problem with achieving super high OCs on the ST was a lack of higher FSB settings than are included in the BIOS. Intel limited the FSB potential of the board. I think a 900MHz OC for 4.1GHz isn't all that bad really. I'm sure the processors can do more. I've heard people attaining 4.2GHz on air, and guess it might have to do with the randomness of individual boards as it does with other components like memory module choice; everything plays a role.

The most I ever accomplished was a 950Mhz OC. I attained a 900MHz OC with my Supermicro and the same chips, but that was without any voltage adjustment, hence my lack of impressment with the ST. In any case, 4GHz and over would be very welcome if I was fortunate enough to have a pair of those Extreme chips. At that frequency, we're looking at well below 30min TPF in -bigadv. My 2.85GHz Harpertowns are doing exactly 36 mIn TPF with an inefficient Linux distro and only 4GB of RAM. So, 4GHz+ would be nice indeed. :cool:

the ST was one of those "omg its an overclockable dual socket motherboard" ploys by intel and it ended up being a bust.. which is what im wondering about with the SR-2..
Can't really say anything about the SR-2 without reading a review or two, but like you I have some doubts. The good thing about the SR-2 is it's designed by EVGA, and they have a very solid enthusiast product line. They know OCing well because it's their business, so despite my reservations for the SR-2's price at launch, I have a feeling it's going to be a decent product. Whether it's worth the high cost of admission is an entirely different debate...

the skulltrail was a beast using LN2 as well but when it came down to water cooling and air it was horrible.. so far we have yet to see any overclocks on the sr-2 besides benchmark LN2 cooling runs..
Another thing we have to take into consideration is dual-socket motherboards were always a hassle to OC. You need two well performing chips and then there's all the additional circuitry that introduces more 'weak links' in addition to the production of more heat. It will always be easier to OC a single processor no matter how you design a board, or what kind of cooling you use with dual sockets.
 
After work tonight - I'll run memtest and see what turns up and also I'll get the BIOS revision as well.

Thanks!
 
i noticed your voltage is .840 seems very low to me for being under load. try 1.0 or 1.2

Memtest is running, almost all done. Will look into the. Oltages, am not sure if I can adjust them or not.

Will be getting the bios rev after memtest is all done? Man this iPhone is hard to typeon
 
Are you using Memtest86+? If so, try using Memtest HCi, apparently it's a much more rigorous test.
 
memtest passed with flying colors and my bios version is 1.0b

Windows Task Manager shows 12gb of RAM

Let me take a look at the event viewer
 
Memtest is running, almost all done. Will look into the. Oltages, am not sure if I can adjust them or not.

Will be getting the bios rev after memtest is all done? Man this iPhone is hard to typeon

hence why i hate the iphone.. a phone is useless without a real keyboard..

but yeah those voltages look insanely low for those clocks.. makes me wonder if theres a lot of vdroop on that motherboard but i would look and see what the board is set at for voltage.. if it really is set to run .89v it makes me wonder what the IMC(NB) is set to run at which could be causing the instability as well which running memtest wont actually pick up because you need the cpu and ram under load at the same time..

i hate seeing prime95 being suggested for stability but in my opinion its really the only way to test this.. run small FFT's for a few hours.. then run large FFT's and see which one crashes.. if crashes during the small then its the cpu and if it crashes during the large then its the ram or NB voltage..
 
hence why i hate the iphone.. a phone is useless without a real keyboard..

but yeah those voltages look insanely low for those clocks.. makes me wonder if theres a lot of vdroop on that motherboard but i would look and see what the board is set at for voltage.. if it really is set to run .89v it makes me wonder what the IMC(NB) is set to run at which could be causing the instability as well which running memtest wont actually pick up because you need the cpu and ram under load at the same time..

i hate seeing prime95 being suggested for stability but in my opinion its really the only way to test this.. run small FFT's for a few hours.. then run large FFT's and see which one crashes.. if crashes during the small then its the cpu and if it crashes during the large then its the ram or NB voltage..

No ways to adjust them voltages...

They're all locked and automatically done.

photo.jpg
 
Wow fuck, I almost updated the wrong BIOS for the wrong motherboard...

My mobo is this

SUPERMICRO MBD-X8DAL-i-O Dual LGA 136
 
Watching eagerly, as my Q6600 system is doing something similar - I swapped out a few things, set it to all stock, and still get a BSOD every 2 or 3 days (down from 2x/day), similar to you. Good luck!

Once you sort out your issue, I might high jack your thread... :D
 
I have also heard reports of some supermico boards having issues with 2xGPUs.
Not sure if this applies to you or not.

Still need to update your bios to the latest one to help rule that out as well.
Here is the link to your board's bios based on the model you stated in post #25.
http://www.supermicro.com/about/policies/disclaimer.cfm?url=/support/resources/getfile.aspx?ID=819

Another diag suggestion is to take out out current video cards are replace with a low power one to see if it is a ps / gpu issue or not.
If you system remains stable, that helps to narrow it down.

If not stable, you are def looking at a cpu / motherboards / or psu issue.
The only other option is to have another user test your CPUs in their board, but i have a very hard time believing it is a cpu issue.
Your load voltages do look very low compared to mine.
X5550Turbo.gif
 
Last edited:
if none of EA's suggestions work.. maybe try taking out 3 sticks of ram on each cpu and see what happens or set the ram to 1066 instead of 1333.. but i agree those voltages look way to low..
 
Back
Top