Subject: Re: DEFZA PMAF-A support
To: Ken Wellsch <kwellsch@tampabay.rr.com>
From: Chris Rupnik <chrisr@beosppc.org>
List: port-pmax
Date: 07/03/2001 10:23:03
Getting better all the time!

alphapci# sysctl -w net.inet.tcp.sendspace=70000
net.inet.tcp.recvspace=70000
net.inet.tcp.sendspace: 16384 -> 70000
net.inet.tcp.recvspace: 16384 -> 70000
alphapci# /usr/pkg/bin/ttcp -t -s alphatc
ttcp-t: buflen=8192, nbuf=2048, align=16384/0, port=5001 tcp -> alphatc
ttcp-t: socket
ttcp-t: connect
ttcp-t: 16777216 bytes in 4.83 real seconds = 3393.45 KB/sec +++
ttcp-t: 2048 I/O calls, msec/call = 2.41, calls/sec = 424.18
ttcp-t: 0.0user 4.1sys 0:04real 87% 0i+0d 0maxrss 0+1pf 0+54csw
alphapci#
alphapci#
alphapci# /usr/pkg/bin/ttcp -t -s -b 65536 alphatc
ttcp-t: buflen=8192, nbuf=2048, align=16384/0, port=5001, sockbufsize=65536
tcp
-> alphatc
ttcp-t: socket
ttcp-t: sndbuf
ttcp-t: connect
ttcp-t: 16777216 bytes in 4.44 real seconds = 3686.49 KB/sec +++
ttcp-t: 2048 I/O calls, msec/call = 2.22, calls/sec = 460.81
ttcp-t: 0.0user 3.8sys 0:04real 87% 0i+0d 0maxrss 0+1pf 0+50csw
alphapci# /usr/pkg/bin/ttcp -t -s -b 65546 -n 4096 alphatc
ttcp-t: buflen=8192, nbuf=4096, align=16384/0, port=5001, sockbufsize=65546
tcp
-> alphatc
ttcp-t: socket
ttcp-t: sndbuf
ttcp-t: connect
ttcp-t: 33554432 bytes in 8.95 real seconds = 3659.95 KB/sec +++
ttcp-t: 4096 I/O calls, msec/call = 2.24, calls/sec = 457.49
ttcp-t: 0.0user 7.6sys 0:08real 86% 0i+0d 0maxrss 0+1pf 0+98csw


----- Original Message -----
From: "Matt Thomas" <matt@3am-software.com>
To: "Ken Wellsch" <kwellsch@tampabay.rr.com>; "Chris Rupnik"
<chrisr@beosppc.org>
Cc: <port-pmax@netbsd.org>
Sent: Tuesday, July 03, 2001 10:22 AM
Subject: Re: DEFZA PMAF-A support


> At 10:15 AM 7/3/2001 -0400, Ken Wellsch wrote:
> >Maybe some wise soul out there will recommend another benchmark,
> >but I've seen "ttcp" mentioned enough it seems acceptable I think.
> >
> >Anyway, I don't know how to achieve what Matt just mentioned, but
> >if you can do so by specifying a larger buffer at the application
> >level, then maybe it would be interesting to see what happens over
> >a longer period and with different data sizes,
> >
> >         e.g. -l 32768 and/or -n 4096 or -n 8192 etc.
>
> -b 65536
>
> sysctl -w net.inet.tcp.sendspace=70000 net.inet.tcp.recvspace=70000
>
> --
> Matt Thomas               Internet:   matt@3am-software.com
> 3am Software Foundry      WWW URL:
http://www.3am-software.com/bio/matt/
> Cupertino, CA             Disclaimer: I avow all knowledge of this message
>
>