Path: utzoo!attcan!uunet!husc6!mailrus!iuvax!pur-ee!uiucdcs!uxc.cso.uiuc.edu!urbsdc!aglew
From: aglew@urbsdc.Urbana.Gould.COM
Newsgroups: comp.arch
Subject: Re: m88000 benchmarks (and C vs ASM
Message-ID: <28200165@urbsdc>
Date: 20 Jun 88 16:07:00 GMT
References: <12112@ut-sally.UUCP>
Lines: 12
Nf-ID: #R:ut-sally.UUCP:12112:urbsdc:28200165:000:658
Nf-From: urbsdc.Urbana.Gould.COM!aglew    Jun 20 11:07:00 1988


>Gosh!  Does this mean that careful hand-coding can yield a factor of 2
>faster code even on a RISC machine?  I know it means that on a CISC machine
>because I've done better than that myself. (Current record: a factor of 22
>on a Nova computer, me vs. Fortran, with the sieve benchmark).

My personal record is a 3-fold improvement -- translating assembly code
into C, and then improving the algorithm. It goes either way.
   I've done better going fom C into assembly, but I'm most proud
of the speedups I've obtained going from assembly into C. I somehow
think that that code is more likely to be running 2 processor 
families from now than the assembly.