>>>>> "Justin" == Justin T Gibbs <gibbs@freefall.freebsd.org> writes:

Justin> I would also suggest using this little program from Bruce
Justin> Evans.  It attempts to measure the combined controller and
Justin> disk command overhead:

Thanks for the little code snippet that computes command overhead and
transfer speed.  However it's very inconsistent in my runs, even when
I raised the iteration count by two magnitudes.  My amiga has seen
command overhead from 1565 us upto 2020 us, and transfer rates from 6
MB/s to 10 MB/s.

This is with my IPL remapping interrupt system which I need for the
ISA support.  Actually that could be the problem with getting
consistent reads, the clockinterrupts might get blocked for longer
than one would want.  Is this test consistent for others?

The fun thing about this is that dd:ing from /dev/zero to /dev/null is
slower, I only get about 5MB/s for that op.  Copying is *expensive* on
the m68k line (esp. with low clock frequencies).  Maybe bcopy.s can do
be written better?

Niklas Hallqvist       Phone: +46-(0)31-40 75 00  Home: +46-(0)31-41 93 95
Applitron Datasystem   Fax:   +46-(0)31-83 39 50  Home: +46-(0)31-41 93 96
Molndalsvagen 95       Email: niklas@appli.se     GSM:  +46-(0)70-714 10 35
S-412 63  GOTEBORG     WWW:   Here
Sweden		       IRC:   niklas (#NetBSD)