On Mon, 14 Mar 2005 18:10, Matt Gibson wrote:
On Monday 14 Mar 2005 02:09 am, Randall R Schulz wrote:
Recently I started getting segfaults when compiling. My memory is relatively new so I tried reseating it; the problem disappeared.
Run a few passes of Memtest86 (<http://www.memtest86.com/>) to get some confidence that your memory is free from errors.
Hear hear! I recently bought two 512Mb memory sticks. I bought Kingston first of all, and one of them turned out to be faulty (but I didn't figure it out for a week because it was the second stick, and I didn't fill up half a gig very often!) I had Corsair as a replacement, and and soon as it arrived, I was paranoid enough to run memtest86 on it. That went back to the seller the same day, because I came back to find several hundred errors reported. The replacement I got for that, exactly the same Corsair, is fine.
Either I was extremely unlucky, or memory's getting a lot less reliable these days. And when the usual symptom that anyone gets is that their Windows machine crashes, I wonder how many people are actually noticing?
The difference between Linux and windows is that Windows only uses the memory when needed by a program whereas Linux any excess for buffering, etc. Quite often you will find a system that runs fine under Windows falls over immediately on Linux for this reason. In my opinion whenever you add memory to a box you should run the memtest86 for at least 12 hours to ensure there are no problems. I have had sticks that only failed once or twice on the memory test in 24 hours but when running Linux the system fell over/locked up every week or so. Quite often the problem occurs when the two RAM sticks have different timings. It is best to use RAM having the same manufacturer and chips or better still only have one stick of double the capacity. -- Regards, Graham Smith ---------------------------------------------------------