🚀 go-pugleaf

On Thu, 10 Apr 2008 21:37:42 -0700 (PDT), penang@freemail.c3.hu put
finger to keyboard and composed:

>I drop XP to safe mode with command prompt, then call up 2 instances
>of the memtest from http://hcidesign.com/memtest/, each checking
>918MB.
>
>Result - Errors !
>
>	Memory Errors Detected !
>	Copying between 4c21c71 and 4c2193a did not result in accurate copy.
>
>	Memory Errors Detected !
>	Pair 19897713 does not store values accurately.
>
>	Memory Errors Detected !
>	Copying between 5547871 and 5547832 did not result in accurate copy.
>
>	Memory Errors Detected !
>	Copying between 4c32471 and 4c323a2 did not result in accurate copy.
>
>	Memory Errors Detected !
>	Copying between 36f8071 and 36f8060a did not result in accurate copy.
>
>	Memory Errors Detected !
>	Copying between 4c32471 and 4c322d9 did not result in accurate copy.
>
>Conclusion:
>	Something wrong with the DDR2 memory module !
>
>Problem:
>	Which module giving out errors?
>
>I got one of the DDR2 module out. Ran the memtest from http://hcidesign.com/memtest/
>-- no error.
>
>Switch the sticks to test the other DDR2 module. Ran the memtest from
>http://hcidesign.com/memtest/ -- again no error !

IME RAM testing software often doesn't properly test refresh. Refresh
errors usually show up in normal operation because, unlike most RAM
diagnostic software, the OS may not read a memory location immediately
after writing it, thus giving the data some time to decay.

See http://www.memtest86.com/tech.html

======================================================================
Test 9 [Bit fade test, 90 min, 2 patterns] 

The bit fade test initializes all of memory with a pattern and then
sleeps for 90 minutes. Then memory is examined to see if any memory
bits have changed. All ones and all zero patterns are used. This test
takes 3 hours to complete. The Bit Fade test is not included in the
normal test sequence and must be run manually via the runtime
configuration menu.
======================================================================

FWIW, I notice that the failing memory addresses all end in "71" which
makes me wonder whether you have an addressing error. It's a pity that
HCI Design didn't see fit to report the bit differences in the failing
data patterns. :-(

- Franc Zabkar
-- 
Please remove one 'i' from my address when replying by email.