lkml.org 
[lkml]   [2004]   [Aug]   [21]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: Possible dcache BUG
Date
On Saturday 21 August 2004 12:25, Barry K. Nathan wrote:
> > Memtest86 may not know howto enable it if its an
> > nforce2 option. Whatever cache shown as switchable in the bios,
> > turning it off makes a very sick bird out of the machine, like a
> > 33mhz 386sx?
>
> Yeah, disabling the L2 cache on a modern CPU makes it really slow. But,
> it's still a useful troubleshooting option...

When I had the problem described in my previous mail I came to the conclussion
that it was related with cache *BUT* it seemed that the cache was just
caching wrong data. Disabling the cache would just reduce the problem.

One reason for this is that when the program detected errors in a buffer (i.e.
0x1234 instead of 0x1111) then they would NOT go away if the program was
reading from this buffer all the time. This means that the cache always
returned the same data. The error was 'gone' every time the program was
suspended for a while or when something else used a lot of memory (i.e.
another instance of this program).
So, I'm not suggesting that his cache is faulty but that there can be a CPU
(or even a M/B) problem that corrupts data when they are transfered from
memory to the processor.

> -Barry K. Nathan <barryn@pobox.com>
<<V13>>
[unhandled content-type:application/pgp-signature]
\
 
 \ /
  Last update: 2005-03-22 14:05    [from the cache]
©2003-2011 Jasper Spaans