[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <1415240055.13896.57.camel@edumazet-glaptop2.roam.corp.google.com>
Date: Wed, 05 Nov 2014 18:14:15 -0800
From: Eric Dumazet <eric.dumazet@...il.com>
To: Rick Jones <rick.jones2@...com>
Cc: David Miller <davem@...emloft.net>,
netdev <netdev@...r.kernel.org>,
Or Gerlitz <ogerlitz@...lanox.com>,
Willem de Bruijn <willemb@...gle.com>
Subject: Re: [PATCH net-next] net: gro: add a per device gro flush timer
On Wed, 2014-11-05 at 17:38 -0800, Rick Jones wrote:
> Speaking of QPS, what happens to 200 TCP_RR tests when the feature is
> enabled?
Nothing at all (but the usual noise I guess)
200 TCP_RR send packets with 1 byte of payload and Push flag,
so no packet ever sits in napi->gro_list
lpaa5:~# echo 0 >/sys/class/net/eth0/gro_flush_timeout
lpaa6:~# echo 0 >/sys/class/net/eth0/gro_flush_timeout
lpaa5:~# time ./super_netperf 200 -H lpaa6 -t TCP_RR -l 20
3.13827e+06
real 0m32.170s
user 0m32.885s
sys 7m38.868s
lpaa5:~# echo 2000 >/sys/class/net/eth0/gro_flush_timeout
lpaa6:~# echo 2000 >/sys/class/net/eth0/gro_flush_timeout
lpaa5:~# time ./super_netperf 200 -H lpaa6 -t TCP_RR -l 20
3.19013e+06
real 0m37.152s
user 0m33.477s
sys 7m30.586s
Now lets try TCP_RR with -- -r 4000,4000 ;)
Reducing ACK packets allow us to better use the 10Gbe bandwith for
payload, so QPS actually increase.
lpaa5:~# echo 0 >/sys/class/net/eth0/gro_flush_timeout
lpaa6:~# echo 0 >/sys/class/net/eth0/gro_flush_timeout
lpaa5:~# time ./super_netperf 200 -H lpaa6 -t TCP_RR -l 20 -- -r
4000,4000
379645
real 0m32.201s
user 0m4.390s
sys 0m59.501s
lpaa5:~# echo 2000 >/sys/class/net/eth0/gro_flush_timeout
lpaa6:~# echo 2000 >/sys/class/net/eth0/gro_flush_timeout
lpaa5:~# time ./super_netperf 200 -H lpaa6 -t TCP_RR -l 20 -- -r
4000,4000
400610
real 0m37.159s
user 0m4.501s
sys 0m59.665s
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists