Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 13 Dec 2001 12:36:09 -0800
From:      Luigi Rizzo <rizzo@aciri.org>
To:        Garrett Wollman <wollman@khavrinen.lcs.mit.edu>
Cc:        bmah@FreeBSD.ORG, net@FreeBSD.ORG
Subject:   Re: -current vs. -stable network performance
Message-ID:  <20011213123608.B41389@iguana.aciri.org>
In-Reply-To: <200112131958.fBDJwef85203@khavrinen.lcs.mit.edu>
References:  <20011212224206.D35108@iguana.aciri.org> <200112131839.fBDId7v70103@apollo.backplane.com> <200112131913.fBDJD8d38312@bmah.dyndns.org> <200112131958.fBDJwef85203@khavrinen.lcs.mit.edu>

next in thread | previous in thread | raw e-mail | index | archive | help
As i said in my original message, I can pump out and receive
almost at full speed using CURRENT, but I can only forward 2/3
of the packets that STABLE does (~80Kpps vs 125Kpps).

The TTCP test is not really significant, as 1.5K packet means
that you are only dealing with less than 10,000 pps, we are talking
one order of magnitude higher here...

	cheers
	luigi

On Thu, Dec 13, 2001 at 02:58:40PM -0500, Garrett Wollman wrote:
> <<On Thu, 13 Dec 2001 11:13:08 -0800, "Bruce A. Mah" <bmah@FreeBSD.ORG> said:
> 
> > 5-CURRENT (11/19):		9244 pps, 35.6 Mbps
> > 4-STABLE (late November):	21827 pps, 84 Mbps
> 
> Doesn't seem right to me.
> 
> wollman@cheyenne-mountain(6)$ ttcp -t -s -v -f m -b 131072 -u mintaka 
> ttcp-t: buflen=8192, nbuf=2048, align=16384/0, port=5001, sockbufsize=131072  udp  -> mintaka
> ttcp-t: socket
> ttcp-t: sndbuf
> ttcp-t: 16777216 bytes in 1.32 real seconds = 97.26 Mbit/sec +++
> ttcp-t: 16777216 bytes in 0.10 CPU seconds = 1276.04 Mbit/cpu sec
> ttcp-t: 2094 I/O calls, msec/call = 0.64, calls/sec = 1591.09
> ttcp-t: 0.0user 0.1sys 0:01real 7% 12i+184d 622maxrss 0+2pf 40+174csw
> ttcp-t: buffer address 0x8054000
> 
> These are 4-stable 1-GHz Pentium IIIs doing UDP on a switched
> 100-Mbit/s network with GA620Ts.  For TCP, it's a little bit worse:
> 
> wollman@cheyenne-mountain(7)$ ttcp -t -s -v -f m -b 131072 mintaka
> ttcp-t: buflen=8192, nbuf=2048, align=16384/0, port=5001, sockbufsize=131072  tcp  -> mintaka
> ttcp-t: socket
> ttcp-t: sndbuf
> ttcp-t: connect
> ttcp-t: 16777216 bytes in 1.55 real seconds = 82.46 Mbit/sec +++
> ttcp-t: 16777216 bytes in 0.11 CPU seconds = 1212.30 Mbit/cpu sec
> ttcp-t: 2048 I/O calls, msec/call = 0.78, calls/sec = 1319.33
> ttcp-t: 0.0user 0.1sys 0:01real 6% 19i+294d 246maxrss 0+2pf 7033+35csw
> ttcp-t: buffer address 0x8054000
> 
> Results are actually a little bit better on my six-months-ago-current
> desktop (an 800-MHz Pentium III with i82559).  These machines are all
> under some level of network load, so the results for a clean system
> should be even better.
> 
> -GAWollman
> 
> 
> To Unsubscribe: send mail to majordomo@FreeBSD.org
> with "unsubscribe freebsd-net" in the body of the message

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-net" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20011213123608.B41389>