Re: Random corruption in atlas computations
On Fri, 15 Oct 2004, Camm Maguire wrote:
> Greetings! The atlas testers fail at random intervals on m68k,
> preventing the package build to complete. I'm suspecting some cache
> or register flushing issue. When I try to stop in gdb at the point
> the corrupted value is supposedly written, the problem never appears.
>
> ...
>
> Ideas?
If you can reproduce the problem on another machine, I'd probably try a
different compiler, and/or try and reduce the test case to a short program
that exhibits the same problem.
If you can't reproduce the problem on another machine, I'd probably
suspect bad RAM. Charles Cazabon's memtester might be able to confirm
that,
http://www.qcc.ca/~charlesc/software/memtester/
-F
Reply to: