[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 1 Jan 2014 18:46:42 +0000
From: Yuval Mintz <yuvalmin@...adcom.com>
To: Amir Vadai <amirv@...lanox.com>,
"David S. Miller" <davem@...emloft.net>
CC: "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
Or Gerlitz <ogerlitz@...lanox.com>,
Eugenia Emantayev <eugenia@...lanox.com>,
Ido Shamay <idos@...lanox.com>
Subject: RE: [PATCH net-next 2/2] net/mlx4: Revert "mlx4: set maximal number
of default RSS queues"
> From: Eugenia Emantayev <eugenia@...lanox.com>
>
> This reverts commit 90b1ebe "mlx4: set maximal number of default RSS
> queues".
> Limiting the number of receive rings to default (8) reduces overall
> packet rate by 15% in multiple TCP streams application:
>
> Packet rate
> 8 rings 967691
> 16 rings 1142070
>
> Results were obtained using netperf:
>
> S=200 ; ( for i in $(seq 1 $S) ; do ( \
> netperf -H 11.7.13.55 -t TCP_RR -l 30 &) ; \
> wait ; done | grep "1 1" | awk '{SUM+=$6} END {print SUM}' )
Seems odd - well, to be more exact, nothing is odd about the results;
the whole notion of the original patch was to set an upper limit on the
number of interrupt vectors multi-queue devices ask by default, not
to improve the default performance.
If you believe this is a better default (or some relaxation will be,
e.g., 16 instead of 8), why not set it as default for ALL multi-queue
networking drivers?
Thanks,
Yuval
--
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