RAM Errors (only during enabled CPU-Load), no errors @MEMTEST 4.1

[0815hero]

Hello,


there is a strange behaviour with my system.
There are some RAM-Errors with BurnIn (Configuration: 100%RAM + 100%CPU).
With the same system there are no errors in MEMTEST 4.1 (more than 500 complete passes!!!)...
With the same system there are (till now) no errors in BurnIn with Configuration RAM 100% (no more tests are selected). This test is still running (more than 4h now)...

It is strange, that at each error the read and expected value are the same and the address are different (details see below).


AND: With some other 2x2GB DDR3 RAMs (e.g. Transcend or Samsung) BurnIn works fine - and MEMTEST too!

Do you have any idea?



My system:


CPU: Penryn 2,26GHz, Core2Duo, SP9300
Chipset: GS45 + ICH9ME
RAM: 2x2GB DDR3, Hynix, HMT125S6AFP8C-G7
Windows 7, 64bit
BurnIn Pro, V6.0, 64 bit

BurnInSettings: 100% CPU Load + 100% RAM Load.

After ~ 2h there are five RAM errors (and no CPU errors!):



Event Log:

SERIOUS: 2010-06-30 11:06:48, RAM, Error verifying data in RAM
LOG NOTE: 2010-06-30 11:06:48, Memory (RAM), Process 1, Data verify error: 0x00000000192E1CF4, read 268435456, expected 0

SERIOUS: 2010-06-30 11:21:55, RAM, Error verifying data in RAM
LOG NOTE: 2010-06-30 11:21:55, Memory (RAM), Process 3, Data verify error: 0x0000000016A355B4, read 268435456, expected 0

SERIOUS: 2010-06-30 11:36:34, RAM, Error verifying data in RAM
LOG NOTE: 2010-06-30 11:36:34, Memory (RAM), Process 2, Data verify error: 0x000000000452A434, read 268435456, expected 0

SERIOUS: 2010-06-30 12:07:19, RAM, Error verifying data in RAM
LOG NOTE: 2010-06-30 12:07:19, Memory (RAM), Process 2, Data verify error: 0x000000000E44E2B4, read 268435456, expected 0

SERIOUS: 2010-06-30 12:08:16, RAM, Error verifying data in RAM
LOG NOTE: 2010-06-30 12:08:16, Memory (RAM), Process 0, Data verify error: 0x000000003095A2B4, read 268435456, expected 0

[David (PassMark)]

We often see cases where memtest fails to detect faults and BurnInTest does. And sometimes we see the other way around as well.

The errors indicate your RAM is bad. In binary the expected value to be read from RAM was,
00000000000000000000000000000
but on your system this value was returned.
10000000000000000000000000000
a single bit difference. Which is typical of RAM faults.

As you have already swapped RAM and the problem disappears, I think this would confirm it is the RAM at fault. If it was a compatibility issue with the MB, or timing issue, or EMI issue, you might still find the same RAM works OK on another MB.

[michelb]

FWIW, I'm also getting RAM errors with BurnInTest but MemTest86+ reports no errors. I do believe the errors are correct and the RAM is bad though as I swapped the memory from another machine, re-ran the BurnInTest and this test was clean. Also after I swapped the RAM, the other machine would not boot into Win7 with the 'bad' RAM. I then re-swapped the RAM back to the way it was originally, ran BurnInTest again on the machine with the bad RAM that was originally getting the errors and got errors again. The 2nd machine, ran fine good with the 'good' RAM.

From this, it does look like I have 1 stick of RAM that's bad and 1 that's good but only BurnItTest reports the error and MemTest86+ reports no errors.