[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bad DIMM? Need testing advice


I recently thought I'd take advantage of the low RAM prices and got some
memory for my home machine (Athlon 900, Asus A7V). I bought two 256MB
PC133-DIMMs specified for 2-2-2 timing.

Being a bit on the cautious side regarding memory, I decided to run
memtest86 which indeed reported some errors early on. I haven't had time
to investigate any further, but I thought I'd post what I have so far and
ask for advioce about my testing plan. Here goes:

During Test 1 [Moving Inv, ones&zeros, cached] a couple of addresses
fairly close to each other (1e92e99c, 1e94a99c, 1e90a99c, 1e92699c,
1e91a99c, 1c93e99c, 1e94299c, 1e91a99c) were reported with errors. They
all seemed to fail in exactly the same bit. The ouput was somewhat as

Addr Good:00000000 Bad:40000000 Xor:40000000

I encountered 8 errors in 7 passes of Test 1, 6 of which occured in pairs
(2 errors in same run, shortly after one another). 2 passes went through
without any errors. As you can see above, no single address failed more
than once.

As I have absolutely no idea about the physical properties of memory, I'm
somewhat at a loss interpreting this result. Does it look like a faulty
chip or could there be some other reason? The memtest86 README mentions
something about USB legacy support producing fake errors with some INTEL
chipsets -- could something similar be happening here? FWIW, I have
disabled USB legacy support.

I plan to proceed as follows:

- run the complete test (could anybody with a similar setup give me some
estimate as to how long this would take? Also, do you deem it necessary to
run the extended tests as well?) at least once, better twice and note
which other tests fail.

- swap the 2 DIMMs and rerun the tests that failed, noting whether the
errors still occur and if yes, whether they occur in the same memory
region. If that is the case, I'd suspect it's not a simple case of a
broken chip.

- test each DIMM separately to find out which one is faulty.

Any comments, suggestions, advice?

Thanks in advance,


Manuel Reiter		 	   |         reiter@th.physik.uni-frankfurt.de
Institut fuer Theoretische Physik  |
J.W.Goethe Universitaet		   |
Robert-Mayer-Str. 8-10		   |
D-60054 Frankfurt am Main	   |
Germany				   |  (Voice: [+49]-69-798-22632, Fax: -28350)

Reply to: