On Mon, Apr 26, 2004 at 11:35:32AM -0700, prashant s wrote:
> sir,
> i have a doubt on NAPI(feature which was
> introduced in linux 2.6.3).In which polling is used
> instead of interrupts when packets arrive at the
> network card.
>
> i did not find any performance difference between a
> non NAPI driver and a NAPI driver(this may be because
> i used large packets).Both of them showed 94Mbps,but
> when i reduced the "DMA receive ring size" from 32k to
> 8k, i got a performance difference. The ordinary
> driver without NAPI (rtl8139too in 2.4.20 linux)
> showed 72Mbps and the NAPI driver (rtl8139too in
> 2.6.3) showed 94Mbps,both having their ring size at
> 8k.The packet size i used was same as before.
>
> why is this difference when the ring size is
> reduced.
>
well, I'm not a guru and I don't know how you made
the measurements but I think that results are quite
reasonables, if you have a queue with a small buffer then
it's more easy to fill it and therefore it's necessary
to drop traffic.
in next days I want to make similar tests with Linux and FreeBSD,
I will post my results.
I would like to know the system, interrupt and ksoftirqd load
in your tests, also the machine processor
cheers
Ulisses
Debian GNU/Linux: a dream come true
-----------------------------------------------------------------------------
"Computers are useless. They can only give answers." Pablo Picasso
Humans are slow, innaccurate, and brilliant.
Computers are fast, acurrate, and dumb.
Together they are unbeatable
---> Visita http://www.valux.org/ para saber acerca de la <---
---> Asociación Valenciana de Usuarios de Linux <---
-
: send the line "unsubscribe linux-net" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
