VOGONS


First post, by 386SX

User metadata
Rank l33t
Rank
l33t

Hello,

I'm trying lately this test config, a Pentium 4 3,6Ghz, an Asus P5 latest bios G31 Socket775 mainboard, second hand 4x 1GB Nanya DDRII-800 rams found cheap and set in Dual Channel and today I found running memtest that both in multithreading and safe mode it find a single bit error on different test at the same address. Is this a sign the module need a replacement or a single bit like that might not be a serious problem beside maybe it might be cheap enough to change it?
Usually when I found memtest dimm errors there were many of them sign that the module was gone I suppose. This one only seem to have that single bit difference. And 406MB is that the first module location?
Thanks for any opinions.

Bye

Last edited by 386SX on 2021-11-28, 11:17. Edited 1 time in total.

Reply 1 of 4, by dionb

User metadata
Rank l33t++
Rank
l33t++

To find out which module it is, test the two separately in single channel. Also, your screenshot shows test running for 16 mins. You generally want to run Memtest overnight. It's possible you'll find many more errors if you do.

Reply 2 of 4, by 386SX

User metadata
Rank l33t
Rank
l33t
dionb wrote on 2021-11-28, 11:16:

To find out which module it is, test the two separately in single channel. Also, your screenshot shows test running for 16 mins. You generally want to run Memtest overnight. It's possible you'll find many more errors if you do.

Thanks, that was the second time it ran (the first in dual thread x64 that took maybe more than 30 minutes) but I'll leave the test running in a long time. 😀

Reply 3 of 4, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

If it's always the same bit at the same address then you can set memtest to test just a few MB around where the fault is to make it a bit faster. 5 errors in 16 minutes does sound like a problem and can lead to random crashes. Some OSes (linux, and I assume others) can use a 'badram' boot parameter to exclude addresses so you can make the computer reliable.

Reply 4 of 4, by 386SX

User metadata
Rank l33t
Rank
l33t

Thanks. I found the single module having the problems testing before two that didn't give any and then one of the others single one and that didn't have errors too. Then tested the three modules together two times almost 2 hours of Pentium 4 Hyperthreading PSU stressing test and no errors on these 3GB of rams. 😁
I'll find another 1GB module just for the faster final bandwidth even if on such cpu it might not make much difference.