Linux MIPS Architecture development
 help / color / mirror / Atom feed
From: Clem Taylor <clem.taylor@gmail.com>
To: "jaypee@hotpop.com" <jaypee@hotpop.com>
Cc: linux-mips <linux-mips@linux-mips.org>
Subject: Re: Au1550 ethernet throughput low
Date: Thu, 14 Jul 2005 20:13:29 -0400	[thread overview]
Message-ID: <ecb4efd1050714171318ce81aa@mail.gmail.com> (raw)
In-Reply-To: <1121354144l.5178l.2l@cavan>

On 7/14/05, jaypee@hotpop.com <jaypee@hotpop.com> wrote:
> Clem you said were using 2.6.11 was that a kernel.org kernel or one
> from linux-mips.org?

The source I'm using originated from http://www.linux-mips.org/. It
was checked out from the head of
':pserver:cvs@ftp.linux-mips.org:/home/cvs' on 2005.03.18. At the time
of checkout, the linux-mips tree was missing a 2.6.11 tag. The closest
tag was linux_2_6_11_rc5, but the code is 2.6.11.

I'm thinking about switching to 2.6.12 next week so I can get PCI
shutdown support.

> Clem sent me the output of a test. In which he is getting 11MBs
> I'm assuming the B is bytes in which case that is as near line rate as dammit.

Yeah, UDP is running at near line rate, but it does consume a bunch of
CPU. I'm running our 1550s at 492MHz, but I have to run the memory at
123MHz (DDR). I just ran the test again, here's what ttcp said:

udp recv on au1550
ttcp -u -r -s -n 16384 -l 32768 -A 32768 -v -b 262144 -f M
ttcp-r: buflen=32768, nbuf=16384, align=32768/0, port=5001,
sockbufsize=262144  udp
ttcp-r: 536608768 bytes in 44.72 real seconds = 11.44 MB/sec +++
ttcp-r: 536608768 bytes in 17.41 CPU seconds = 29.39 MB/cpu sec
ttcp-r: 16378 I/O calls, msec/call = 2.80, calls/sec = 366.21
ttcp-r: 0.1user 17.2sys 0:44real 38% 0i+0d 0maxrss 0+7pf 16344+13csw

udp xmit on au1550
ttcp -u -t -n 16384 -s -l 32768 -A 32768 -v -b 262144 -f M  gort
ttcp-t: buflen=32768, nbuf=16384, align=32768/0, port=5001,
sockbufsize=262144  udp
ttcp-t: 536870912 bytes in 44.76 real seconds = 11.44 MB/sec +++
ttcp-t: 536870912 bytes in 15.26 CPU seconds = 33.56 MB/cpu sec
ttcp-t: 16390 I/O calls, msec/call = 2.80, calls/sec = 366.14
ttcp-t: 0.1user 15.1sys 0:44real 34% 0i+0d 0maxrss 0+8pf 3272+10csw

  reply	other threads:[~2005-07-15  0:12 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2005-07-13 16:00 Au1550 ethernet throughput low jaypee
2005-07-14 15:02 ` Matej Kupljen
2005-07-14 15:06   ` Pete Popov
2005-07-14 15:15   ` jaypee
2005-07-15  0:13     ` Clem Taylor [this message]
2005-07-15  8:21       ` jaypee
2005-07-15  9:17         ` Bruno Randolf
2005-07-18  9:57           ` jaypee
2005-07-18 12:56             ` Dan Malek
2005-07-18 13:41               ` jaypee
     [not found]             ` <ecb4efd105071809082628bb27@mail.gmail.com>
2005-07-18 17:25               ` jaypee
  -- strict thread matches above, loose matches on Subject: below --
2005-07-14  5:02 Singh, Ajay
2005-07-14  5:02 ` Singh, Ajay
2005-07-14 13:52 ` jaypee

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=ecb4efd1050714171318ce81aa@mail.gmail.com \
    --to=clem.taylor@gmail.com \
    --cc=jaypee@hotpop.com \
    --cc=linux-mips@linux-mips.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox