Subject: Re: More on Pentium gcc
To: Jeff Thieleke <>
From: Perry E. Metzger <>
List: port-i386
Date: 04/22/1996 07:10:50
I noticed a lot of the people benchmarking have been using gzip -9;
what are the results like for ordinary gzip? The CPU usage patterns
are different, you know.

I'd also like to hear how nroff does.

BTW, 5-10% reduction in code size and 5-10% improvement in speed is
still very worthwhile.

Jeff Thieleke writes:
> > gzip isn't a good test.  I'll run hint and see how things go, and put
> > graphs up on my web site.
> Possibly, but since the PWG uses gzip as its demonstration of the 
> optimization of gcc, I wanted to see how well it held up.
>     "GZIP-1.24 is 30% faster on unzipping, and with some experimental 
>      version of PGCC, its 35% faster!"
> Since I only saw an improvement of 11% on unzipping, I will be interested 
> on how your benchmark turns out.
> Jeff Thieleke