From owner-freebsd-hackers Sat Dec 23 22:22:47 1995 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id WAA03941 for hackers-outgoing; Sat, 23 Dec 1995 22:22:47 -0800 (PST) Received: from main.gbdata.com (dial37.phoenix.net [199.3.234.72]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id WAA03858 for ; Sat, 23 Dec 1995 22:21:23 -0800 (PST) Received: (from gclarkii@localhost) by main.gbdata.com (8.6.12/8.6.9) id AAA01040; Sun, 24 Dec 1995 00:21:27 -0600 From: Gary Clark II Message-Id: <199512240621.AAA01040@main.gbdata.com> Subject: Re: Pentium bcopy 486DX-66 results To: davidg@Root.COM Date: Sun, 24 Dec 1995 00:21:25 -0600 (CST) Cc: freebsd-hackers@FreeBSD.ORG Reply-To: gclarkii@FreeBSD.ORG In-Reply-To: <199512240541.VAA00518@corbin.Root.COM> from "David Greenman" at Dec 23, 95 09:40:58 pm X-Mailer: ELM [version 2.4 PL24] Content-Type: text Sender: owner-hackers@FreeBSD.ORG Precedence: bulk David Greenman wrote: > > >On Sun, 24 Dec 1995, Torbjorn Granlund wrote: > >> > >> I don't have any 486 systems, so I cannot tell for sure. Here is a simple > >> test program that you can use for timing tests: > > > > here are results using 2.1.0-RELEASE, a 486dx-66 with an asus sp3g > >motherboard and write-back cache discipline. > > > >Aspen:[45] ./a.out > >copy 8259 > >memcpy 8582 > > > > You should have just used cc to assemble it - name the file copy.S and then > use: > > cc -c copy.S > cc -o copytest copytest.c copy.o > > This will run it through cpp. ...thanks for the results. > > -DG Hi, Here are my results on a 486DX4-100 8 meg in X not sure about cache. gcc -m486 -O2 -o testcopy testing.c copy.o main:gclarkii# time testcopy copy 10232 bcopy 2403 memcpy 9315 21.9u 0.0s 0:22.08 99.5% 5+183k 1+0io 0pf+0w My sources are using 1024 for SIZE and I just change copy for bcopy for the bcopy stats. It looks like copy is a big loss for me...:( Gary Gary Clark II (N5VMF) gclarkii@FreeBSD.ORG