> Hi,
>
> On Sat, 26 Jun 1999 18:57:48 -0700 (PDT), Rudolf Leitgeb
> <leitgeb@variable.stanford.edu> said:
>
> > 2.2.10:
> > First test run: 285s user, 26.58s system, 3:30 elapsed
> > 2.3.8:
> > First test run: 298s user, 29.8s system, 3:44 elapsed
>
> > I also observed the following: While I did run into swap when I ran
> > the test under 2.2.10, swapping did not occur under 2.3.8. This makes
> > it even more surprising that 2.3.8 came out slower than 2.2.10,
> > especially after the announcement, that 2.3.8 came with speed
> > improvements for SMP systems.
>
> No, it sounds as if most of the difference is coming from the fact that
> 2.2.10 _did_ swap, and so had more memory to play with when it camt to
> doing useful work. Tuning this right is still an ongoing work for 2.3.
>
> --Stephen
Well, I certainly _can_ tell you that 2.3.8 was unbelievably faster than
2.2.10 in the benchmarks that I'm running. It's not that I'm memory starved
(1 GB), just _really_ heavy on filesystem activity. Although from what I see
so far (hoping 2.3.9final fixes this) we really hit a brick wall at a certain
point, and filesystem performance drops right back to where it was with
2.2.10, if not lower. But that's way past a hundred users banging the heck
out of the filesystem, so I don't know how often that case is tested...
- Pete
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/