From: Eric Dumazet <eric.dumazet@gmail.com>
To: Vijay Subramanian <subramanian.vijay@gmail.com>
Cc: David Miller <davem@davemloft.net>, netdev@vger.kernel.org
Subject: Re: [RFC] tcp: use order-3 pages in tcp_sendmsg()
Date: Thu, 20 Sep 2012 07:37:55 +0200 [thread overview]
Message-ID: <1348119475.31352.60.camel@edumazet-glaptop> (raw)
In-Reply-To: <CAGK4HS_ZKKrzFh+c3WDxbkMFpDTg7ehw9ydp39k6JPFA_j1YOw@mail.gmail.com>
On Wed, 2012-09-19 at 15:20 -0700, Vijay Subramanian wrote:
> > I did some tests and got no problem so far, even using splice() [ this
> > one was tricky because it only deals with order-0 pages at this moment ]
> >
> > NIC tested : ixgbe, igb, bnx2x, tg3, mellanox mlx4
>
>
> I applied this patch to net-next and tested with e1000e driver.
> With iperf I got around 8 % improvement on loopback.
>
> Tested-by: Vijay Subramanian <subramanian.vijay@gmail.com>
>
>
> Vijay
If you keep the producer and consumer on separate cpus, and use large
enough send() (64KB or 128KB), gain is more like 15 or 20%
iperf uses 8KB writes, while netperf uses a 16KB default.
TCP stack has a problem because /proc/sys/net/ipv4/tcp_reordering
default value (3) is too small for loopback, since a packet contains 4
MSS
A single reorder and some packets are retransmitted.
Following setting is better
echo 16 >/proc/sys/net/ipv4/tcp_reordering
loopback is lossless, so its always surprising we can have TCP
retransmits on this medium ;)
Thanks
next prev parent reply other threads:[~2012-09-20 5:38 UTC|newest]
Thread overview: 37+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-09-17 7:49 [RFC] tcp: use order-3 pages in tcp_sendmsg() Eric Dumazet
2012-09-17 16:12 ` David Miller
2012-09-17 17:02 ` Eric Dumazet
2012-09-17 17:04 ` Eric Dumazet
2012-09-17 17:07 ` David Miller
2012-09-19 15:14 ` Eric Dumazet
2012-09-19 17:28 ` Rick Jones
2012-09-19 17:55 ` Eric Dumazet
2012-09-19 17:56 ` David Miller
2012-09-19 19:04 ` Alexander Duyck
2012-09-19 20:18 ` Ben Hutchings
2012-09-19 22:20 ` Vijay Subramanian
2012-09-20 5:37 ` Eric Dumazet [this message]
2012-09-20 17:10 ` Rick Jones
2012-09-20 17:43 ` Eric Dumazet
2012-09-20 18:37 ` Yuchung Cheng
2012-09-20 19:40 ` David Miller
2012-09-20 20:06 ` Rick Jones
2012-09-20 20:25 ` Eric Dumazet
2012-09-21 15:48 ` Eric Dumazet
2012-09-21 16:27 ` David Miller
2012-09-21 16:51 ` Eric Dumazet
2012-09-21 17:04 ` David Miller
2012-09-21 17:11 ` Eric Dumazet
2012-09-23 12:47 ` Jan Engelhardt
2012-09-23 16:16 ` David Miller
2012-09-23 17:40 ` Jan Engelhardt
2012-09-23 18:13 ` Eric Dumazet
2012-09-23 18:27 ` David Miller
2012-09-20 21:39 ` Vijay Subramanian
2012-09-20 22:01 ` Rick Jones
2012-11-15 7:52 ` Yan, Zheng
2012-11-15 13:06 ` Eric Dumazet
2012-11-16 2:36 ` Yan, Zheng
2012-11-15 13:47 ` Eric Dumazet
2012-11-21 8:05 ` Yan, Zheng
2012-11-15 18:33 ` Rick Jones
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=1348119475.31352.60.camel@edumazet-glaptop \
--to=eric.dumazet@gmail.com \
--cc=davem@davemloft.net \
--cc=netdev@vger.kernel.org \
--cc=subramanian.vijay@gmail.com \
/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