[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 10 Oct 2019 19:16:52 +0100
From: Edward Cree <ecree@...arflare.com>
To: Alexander Lobakin <alobakin@...nk.ru>,
"David S. Miller" <davem@...emloft.net>
CC: Jiri Pirko <jiri@...lanox.com>, Eric Dumazet <edumazet@...gle.com>,
"Ido Schimmel" <idosch@...lanox.com>,
Paolo Abeni <pabeni@...hat.com>,
Petr Machata <petrm@...lanox.com>,
Sabrina Dubroca <sd@...asysnail.net>,
Florian Fainelli <f.fainelli@...il.com>,
Jassi Brar <jaswinder.singh@...aro.org>,
"Ilias Apalodimas" <ilias.apalodimas@...aro.org>,
<netdev@...r.kernel.org>, <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH net-next 2/2] net: core: increase the default size of
GRO_NORMAL skb lists to flush
On 10/10/2019 15:42, Alexander Lobakin wrote:
> Commit 323ebb61e32b ("net: use listified RX for handling GRO_NORMAL
> skbs") have introduced a sysctl variable gro_normal_batch for defining
> a limit for listified Rx of GRO_NORMAL skbs. The initial value of 8 is
> purely arbitrary and has been chosen, I believe, as a minimal safe
> default.
8 was chosen by performance tests on my setup with v1 of that patch;
see https://www.spinics.net/lists/netdev/msg585001.html .
Sorry for not including that info in the final version of the patch.
While I didn't re-do tests on varying gro_normal_batch on the final
version, I think changing it needs more evidence than just "we tested
it; it's better". In particular, increasing the batch size should be
accompanied by demonstration that latency isn't increased in e.g. a
multi-stream ping-pong test.
> However, several tests show that it's rather suboptimal and doesn't
> allow to take a full advantage of listified processing. The best and
> the most balanced results have been achieved with a batches of 16 skbs
> per flush.
> So double the default value to give a yet another boost for Rx path.
> It remains configurable via sysctl anyway, so may be fine-tuned for
> each hardware.
I see this as a reason to leave the default as it is; the combination
of your tests and mine have established that the optimal size does
vary (I found 16 to be 2% slower than 8 with my setup), so any
tweaking of the default is likely only worthwhile if we have data
over lots of different hardware combinations.
> Signed-off-by: Alexander Lobakin <alobakin@...nk.ru>
> ---
> net/core/dev.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/net/core/dev.c b/net/core/dev.c
> index a33f56b439ce..4f60444bb766 100644
> --- a/net/core/dev.c
> +++ b/net/core/dev.c
> @@ -4189,7 +4189,7 @@ int dev_weight_tx_bias __read_mostly = 1; /* bias for output_queue quota */
> int dev_rx_weight __read_mostly = 64;
> int dev_tx_weight __read_mostly = 64;
> /* Maximum number of GRO_NORMAL skbs to batch up for list-RX */
> -int gro_normal_batch __read_mostly = 8;
> +int gro_normal_batch __read_mostly = 16;
>
> /* Called with irq disabled */
> static inline void ____napi_schedule(struct softnet_data *sd,
Powered by blists - more mailing lists