Subject: Re: Any caveats on non-Pentium CPU's?
To: Simon Burge <simonb@wasabisystems.com>
From: Kent Polk <kent@goathill.org>
List: port-i386
Date: 07/19/2001 22:39:09
Simon Burge wrote:
> Kent Polk wrote:
>
> > On 19 Jul 2001 18:35:05 -0500, Rick Kelly wrote:
> > >
> > > speedy# /usr/bin/time dd if=/dev/zero of=/dev/null bs=64k count=16384
> > > 16384+0 records in
> > > 16384+0 records out
> > > 1073741824 bytes transferred in 1 secs (1073741824 bytes/sec)
> > > 0.53 real 0.00 user 0.52 sys
> >
> > and I get:
> >
> > $ /usr/bin/time dd if=/dev/zero of=/dev/null bs=64k
> > count=16384
> > 16384+0 records in
> > 16384+0 records out
> > 1073741824 bytes transferred in 1 secs (1073741824 bytes/sec)
> > 0.39 real 0.01 user 0.37 sys
> >
> > Why would it get exactly the same transfer rate? Appears something
> > on the mb maxed out unrelated to cpu speed.
>
> Note that 64k*16384 = 1073741824. Also, your real time is 0.39 seconds
> and Rick's is 0.53 seconds. It appears that dd(1) only divides by the
> number of (partial?) seconds. Maybe count=1638400 (100 times longer)
> would provide a more usable benchmark..
Pffft. I saw that and promptly discarded it. :^( This looks a lot
better:
$ /usr/bin/time dd if=/dev/zero of=/dev/null bs=64k count=1638400
1638400+0 records in
1638400+0 records out
107374182400 bytes transferred in 39 secs (2753184164 bytes/sec)
39.51 real 0.98 user 38.30 sys