Error Verifying Data in RAM

[Coops]

I am just curious really as to if this could or could not be a fault. Sometimes when we run a unit on Burn Test v4.0 Build 1030 for a long time. Such as if we leave some units running over night in the morning the unit will have failed with 2 or 3 ram errors. However the actually program has performed some where in the region of 20000 cycles of the Ram test.
Further testing from various other programs such as memtest86 and PC Check do not show anything wrong with the Ram.
So the question is, is the ram actually faulty. It does not happen on every unit. Most pass, but a handful will have a couple of errors in a huge amount of successful cycles. Could it just be a glitch in such an extended testing period.

[David (PassMark)]

Most RAM errors are really RAM errors. There is a small chance however that it could be the bus at fault (e.g. to much electrical noise corrupting data) or the CPU at fault.

You should swap the RAM and re-test to make sure it was the RAM at fault.

Could it just be a glitch in such an extended testing period

Maybe it was a "glitch", but isn't this exactly what you are trying to detect and fix. Nobody whats a computer that has glitches.

One glitch per month might be acceptable for a home user. But how happy would you be if you found out that your bank had computers that glitched serveral times a day causing the amount of cash in your bank account to vary at random.

Of course if the glitch add a few zero to balance I guess you would be pretty happy

---
David