Re: Linux 2.4.5-ac2

Leeuw van der, Tim (tim.leeuwvander@nl.unisys.com)
Mon, 28 May 2001 10:19:01 -0500


Alan Cox wrote:

> > Performance is back to that of 2.4.2-ac26, and stability is a lot >
better. Under
> > heavy FS pressure 2.4.5-ac2 is about 5-10% faster than vanilla 2.4.5, >
the aa1,2
> > kernels have the same performance of vanilla 2.4.5.
> >
> > Which one of your changes affected performance so much?

> Its much more a case that the 2.4.5 tree got fixed and I picked up the >
2.4.5
> changes. Its still not perfect (bigmem will deadlock again as in 2.4.5 >
vanilla
> now) but its a much better basis to work from again

But the claim was that 2.4.5-ac2 is faster than 2.4.5 plain, so which
changes are in 2.4.5-ac2 that would make it faster than 2.4.5 plain? Also, I
don't know if 2.4.5-ac1 is as fast as 2.4.5-ac2 for Fabio. If not, then it's
a change in the 2.4.5-ac2 changelog. If it is as fast, it is one of the
changes in the 2.4.5-ac1 changelog.

I haven't tried plain kernels for a long time, but 2.4.5-ac1 is definately a
lot slower for interactive use than 2.4.4-ac8 so I've gone back to using
that version. I haven't tried 2.4.5-ac2 to see if that improves upon -ac1.

The VM in 2.4.5 might be largely 'fixed' and I know that the VM changes in
-ac were considered to be but still broken, however for me they worked
better than what is in 2.4.5.

I have a rather aging P5MMX at 200MHz with 64MB RAM, and I'm only judging
interactive use (not measuring anything like compile times etc).

with greetings,

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