Possible solutions where memory is good but memory errors occur anyway

Not open for further replies.


Posts: 14   +0
After having a busy couple of months building computers for friends and others, I noticed that ‘after construction’ problems seemed to centre around computers that had two sticks of memory.

The memory I installed was the same (two sticks ordered at the same time eg two 512meg sticks). In one case, not yet knowing that the memory may be to blame and going through the endless process of component substitution, I replaced the two 512meg sticks with a matched pair from my own computer (my computer has dual channel RAM and I had two matched sets installed, 2 gig total).

Problem solved. I also noticed that memory read errors disappeared from my home computer at the same time (windows server 2003/gigabyte 8KNXP Ultra 64/3.2 gig Prescott).

I had also noted that memory read errors never occur after my computer has been switched on for a day or two eg it has been constantly on for weeks now and no memory read errors after the first few hours.

So here is my conclusion – for single channel RAM, always get matched pairs anyway and enjoy a crash free environment;
For dual channel with more than one set of RAM, try swapping one pair. Alternatively, try leaving the computer on continuously. You can switch the monitor off and leave the computer idling, or better still, donate your spare cycles to a good cause (eg SETI).

In Sum:
The faster the RAM, the more critical the matching even in single channel systems. Leaving the computer on may in itself facilitate matching for whatever reason (eg components eventually reach the same temperature).

Kind regards,
Robert Karl Stonjek
Not open for further replies.