Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 3 Dec 1996 00:45:12 -0500
From:      "David S. Miller" <davem@jenolan.rutgers.edu>
To:        mark@quickweb.com
Cc:        jgreco@brasil.moneng.mei.com, jkh@time.cdrom.com, dyson@freebsd.org, dennis@etinc.com, kpneal@pobox.com, hackers@freebsd.org, torvalds@cs.helsinki.fi, lm@engr.sgi.com, iain@sbs.de, sparclinux@vger.rutgers.edu
Subject:   Re: TCP/IP bandwidth bragging
Message-ID:  <199612030545.AAA18646@jenolan.caipgeneral>
In-Reply-To: <Pine.BSF.3.94.961203001921.29153A-100000@vinyl.quickweb.com> (message from Mark Mayo on Tue, 3 Dec 1996 00:28:17 -0500 (EST))

next in thread | previous in thread | raw e-mail | index | archive | help
   Date: Tue, 3 Dec 1996 00:28:17 -0500 (EST)
   From: Mark Mayo <mark@quickweb.com>

   Just curious, what was the benchmark you ran? I can't remember it being
   referenced in the thread..

   I'd like to run it on a few machines (ranging from DEC Unix, to Ultrix..)
   and see how my machines are performing. lmbench seems to be the implied
   benchmark, but I'd like to know for sure  =)

Yes, it is lmbench that I have been running most of the time.  I use
"ttcp" once in a while as well, that can be obtained from:

ftp.sgi.com:/sgi/src/ttcp  (I think thats it, you'll have to rummage around)

   I'll give it a run on the 100MB/s net here, and the FDDI. Of course, the
   PC's won't have the bus bandwidth to sustain transfer across the ATM
   switch - but I'd like to make my own comparisons, whether the benchmark
   represents the _real world_ of not is of no concern to me really. I know
   how the machines perform during normal operation, I'm just curious about
   how the benchmark will vary form OS to OS and from hardware to hardware!

Note that with PCI bus theoretically it can be made to keep with an
ATM interface with lots of operating system tricks such as page
flipping.

For FDDI, the latency numbers might not be so hot, depending upon how
low you have the "Token Hold Time" configured on all the cards on your
ring.  And not that although a low token hold-time improves latencies
(many small quick transfers) it will hurt bandwidth and thus a high
token hold-time is recommended for high bandwidth usage.

---------------------------------------------////
Yow! 11.26 MB/s remote host TCP bandwidth & ////
199 usec remote TCP latency over 100Mb/s   ////
ethernet.  Beat that!                     ////
-----------------------------------------////__________  o
David S. Miller, davem@caip.rutgers.edu /_____________/ / // /_/ ><



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199612030545.AAA18646>