Could you try the same test, but sending to /dev/null - to
	see what the numbers are like without disk access speeds.

		David/absolute

                      Neighbours should be one of:
                a) Understanding   b) Distant   c) Deaf

On Tue, 15 Sep 1998, Jesper Brondum wrote:

> >Jesper Brondum wrote:
> >> 
> >> Hi
> >> 
> >> I made a test (in life test) of 3 unix's to a ss1+
> >> 
> >> ftp a 12MB file to the ss1+ (with identical disks)
> >> 
> >> SunOS 4.1.3    699 kb/s    (8 meg ram)
> >> NetBSD 1.3.2   420 kb/s    (16 meg ram)
> >> S/Linux RH 5.1 510 kb/s    (16 meg ram)
> >> 
> >> How is it that the difference is that big ...
> >
> >
> >  You seem to have changed the machine configuration (RAM) between
> >tests.  Did the 4.1.3 test use a SCSI/Ethernet (FSBE) combo Sbus card? 
> >The SS1, 1+, IPC, and SLC have onboard SCSI & Ethernet that share a DMA
> >channel.  The ethernet has a higher priority, so when there's high
> >network activity, the disk I/O rate on the onboard SCSI interface drops.
> >  There are several other questions that need to be answered... Was the
> >other host system changed?  Was the other host idle during all three
> >tests?  How about the ethernet segment the transfer occured over?  Did
> >you make any router hops?  Etc... Etc...
> 
> All three systems are identical, exept that the SunOS machine has less mem.
> 
> (They share the same HUB ... )
> 
> 
> --------------------------------------------------------------------------
>                       Jesper Brondum, Cognitive AB
>                             mailto:jb@ct.se
>         telefon 046-130976, telefax 046-130975, mobil 0706-277766
> 
>