Re: NUMA: Patch for node based swapping

From: Andi Kleen
Date: Tue Oct 12 2004 - 15:00:59 EST


Rik van Riel <riel@xxxxxxxxxx> writes:

> On Tue, 12 Oct 2004, Christoph Lameter wrote:
>
>> The minimum may be controlled through /proc/sys/vm/node_swap.
>> By default node_swap is set to 100 which means that kswapd will be run on
>> a zone if less than 10% are available after allocation.
>
> That sounds like an extraordinarily bad idea for eg. AMD64
> systems, which have a very low numa factor.

As a optional sysctl it makes sense even on AMD64. On some benchmarks
you see the differences between local and remote memory very clearly.

-Andi

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/