Re: On linux kernel packaging issue

[Michael Poole <mdpoole@troilus.org>]

Eduard Bloch writes:

> Do you see now that 8 of your 10 percent come directly from the
> application code and other two maybe from the optimized libc? There is
> not{hing| much} we have won using an optimised kernel. But the placebo
> effect has been demonstraded once again.

You have not shown what you claim you have shown.  You have shown that
for one specific application you get a 10% speedup when you optimize
user-space (which, like I said before, is where it spends almost all
its time).  Martin Schulze suggested elsewhere in this thread that 10%
is where you should start being interested in optimizating user space,
and that the kernel may see useful benefit from lower speedups.

bzip2 is not representative of applications that spend significant
amounts of time in the kernel, but people do run such applications and
spend lots of CPU time in the kernel.  If you want to argue about
optimizing the kernel, find a benchmark that measures the kernel's
performance.  As with many media players, Linux changes more than just
gcc's code generation when you target specific processors.  You cannot
measure its performance by staying in user space like bzip2 does.

Michael