> > Lars Marowsky-Bree <lmb@pointer.teuto.de> writes:
> >
> > > is egcs capable of compiling a working 2.0.32 kernel?
> >
> > I compiled the 2.0.32 kernel I'm currently running on with a
> > snapshot of egcs prior to 1.0 release + pentium gcc patches.
> >
> > $ uptime
> > 11:44pm up 19 days, 4:43, 15 users, load average: 0.06, 0.10, 0.20
> >
> > > Is there any speed difference?
> >
> > Yes! Pgcc speeds up XEmacs very noticeably at -O6, I compiled the
> > kernel at a more modest optimization level (-O2), but everything
> > seems faster. [...]
>
> I've been using pgcc as my primary compiler for several months without problems. Many parts of my system (kernel, libc, binutils and many applications) are pentium optimized. I generally use -O6 -mpentium options (including compilation of kernels). I haven't noticed any kernel oops, or
> anything like this (in stable kernels).
>
> Unfortunately I can't give data about speed difference. I haven't measured it. But some applications' speed increases "visually" - no need to measure it ;)
I too have been running a full pgcced system for quite some time... It's
broken a few apps in the past (like several months ago) but it seems to
have reached a point of stability.. So I've just recompiled everything
with a current release..
Using unixbench I have measured a 6% performance increase when just the
kernel is pgcced.. (I think that compiling the benchmark with pgcc brought
on a whopping 25% increase...)
Some apps it's big.. Some it's small.. But it's never a degergration..