On Wed, Aug 08, 2001 at 07:31:18PM -0400, Jeff Lane wrote:
>
> On Tue, 7 Aug 2001 [EMAIL PROTECTED] wrote:
> > A possible workaround for the speed issue would be to try booting with
> > lilo arg "mem=128M" and see if having less memory helps.
>
> Actually he has 8 gigs of ram... thats a lot. the stock kernel doesnt
> work well with that much (I think the default is 1 or 2 gig).
>
> Recompile your kernel, and turn on bigmem so that the kernel can see and
> use up to 64gigs of ram. then you should be set.
>
> Setting mem=128M just cripples your machine. Sure it boots faster now
> than before BUT that was because you had more ram than your current kernel
> could handle. so recompile and make sure that bigmem is set correctly and
> then you will see your big beefy machine scream like a .. oh well, you get
> the idea..
Perhaps I should have been more explicit and said "temporary workaround"
instead of just "workaround". My point was that mem=128M is a suitable
workaround to allow decent performance while trying to solve this person's
large memory performance problem (such as by building a new kernel).
I certainly would not suggest that someone with 8GB run with 128MB as
other than a temporary measure.
As I'm sure you know, more recent 2.4 kernels have substantial
improvements in their performance for large memory configurations
as compared to earlier 2.4 kernels, which had problems.
As the original poster later wrote:
> OK. I compiled a 2.4.7 kernel from kernel.org. This has totally
> resolved the performance issues.
Also, unless I'm mistaken CONFIG_BIGMEM is not present in
2.4 kernels. It was present in 2.2 kernels.
_______________________________________________
Seawolf-list mailing list
[EMAIL PROTECTED]
https://listman.redhat.com/mailman/listinfo/seawolf-list
