New Dual Xeon 3.4Ghz, DDRII system reboots in 10 minutes

[Anonymous]

Here is my system config.
Supermicro X6DH8-G2 motherboard.
1GB DDR2 400mhz memory
dual 3.4Ghz Xeon CPU 800mhz FSB
IDE 250GB drive
Running windows 2003 enterprise server

The machine reboot with burnin 4.0 Pro and 3.2 pro. No crash message, simply reboots the machine.
I am selecting 100% stress test on CPU, MMX, Memory, HD, and network.

The machine seen to run fine under burnin 2.2 PRO
(still testing, its been running for 45 minutes.)

Thanks for any assistance.

[David (PassMark)]

This is not a known problem.

We have a similar report of a crash (not a reboot) on a Dual 3.2 Xeon machine, and believe that it is related to the particular video card in use. Which is the NVIDIA Quadro FX 3400. Becuase these new Dual Xeon machines have the new PCI-E bus for video cards there are a limited number of video cards available. Maybe you have the same card?

However becuase you say that you are not using the 3D test I think your problem might be different. As the machine reboots the problem is not with BurnInTest, more likely it is hardware or a buggy device driver. If it was a BurnInTest bug, then you would expect BurnInTest to crash and the machine not to turn off.

Did you check the Windows event log for strange messages.

Did you try running just one test type at a time to try and narrow down the fault.

Are you sure your power supply is big enough.

Is the machine stable the rest of the time?

--------
David
PassMark

[stever]

Reboot of xeon 800 on supermicro = bad cpu?

I had this issue also, very rarely it would report memory errors but usually either freeze or reboot.

After replacing memory 3x, disks, raid carriers, cables, and eventually the entire server was replaced - None of which fixed it.

The CPU supplier suggested replacing the CPUs with retail box ones - Instead of OEM Xeons and that fixed the problem.

The chips were OEM, they were the same stepping code, they were 800 mhz, etc. Not wrongly labelled / sold wrong, they seemed to be 100% correct and 100% the same as the replacements, just OEM instead of Retail box.

I should add there was no heat problem with the CPUs, in fact the box ran quite cool.

The retail cpus fixed the problem, box has been 100% stable since.

NOTE: Either CPU by iteself wold run for 24 hours without failure on any test we tried, passmark included.

NOTE 2: Server passed memtest86 for 3 days with both cpus in it, but windows would crash it within 15 minutes of burn in test.

There were a few other stability issues, I don't think this was a passmark issue.

I can't find anything that says the 800 mhz / emt64 OEM cpus are only able to be used one per box, but I know a Xeon MP is needed for > 2 cpus, and AMD Opterons have the 100 versus 200 series for single cpu versus dual compatible - Perhaps there's an issue with Xeon OEMs that I couldn't find.

Or perhaps I had just very odd problems with the CPUs that only showed up with 2 cpus in the box - This seems highly unlikely however.