Stephen Frost <sfrost@snowman.net> writes:
> * Greg Stark (gsstark@mit.edu) wrote:
>> However I was under the impression that 2.6 had moved beyond that problem.
>> It would be very interesting to know if 2.6 still suffers from this.
> The tests on the em64t at my place were using 2.6.12. I had thought 2.6
> was better about this too, but I don't have another explanation for it.
The 4-way Opteron I've been using at Red Hat is running
2.6.12-1.1398_FC4smp (Fedora Core 4 obviously). Red Hat in particular
has been working hard in this area, and I thought that their recent
kernels included NUMA fixes that weren't yet accepted upstream (at least
not in the stable kernel branches). But it seems there's still a ways
to go yet.
It'd be real interesting to see comparable numbers from some non-Linux
kernels, particularly commercial systems like Solaris.
regards, tom lane