[Top] [All Lists]

Re: LLTX and netif_stop_queue

To: hadi@xxxxxxxxxx
Subject: Re: LLTX and netif_stop_queue
From: Roland Dreier <roland@xxxxxxxxxxx>
Date: Mon, 03 Jan 2005 07:57:14 -0800
Cc: Eric Lemoine <eric.lemoine@xxxxxxxxx>, Andi Kleen <ak@xxxxxxx>, Patrick McHardy <kaber@xxxxxxxxx>, "David S. Miller" <davem@xxxxxxxxxxxxx>, netdev@xxxxxxxxxxx, openib-general@xxxxxxxxxx
In-reply-to: <1104764660.1048.578.camel@xxxxxxxxxxxxxxxx> (jamal's message of "03 Jan 2005 10:04:21 -0500")
References: <52llbwoaej.fsf@xxxxxxxxxxx> <20041217214432.07b7b21e.davem@xxxxxxxxxxxxx> <1103484675.1050.158.camel@xxxxxxxxxxxxxxxx> <5cac192f04122210491d64d4b6@xxxxxxxxxxxxxx> <20041222202919.057b8331.davem@xxxxxxxxxxxxx> <5cac192f0412230110628749e3@xxxxxxxxxxxxxx> <41CAF444.3000305@xxxxxxxxx> <5cac192f04122408102129af43@xxxxxxxxxxxxxx> <1104240717.1100.66.camel@xxxxxxxxxxxxxxxx> <5cac192f0501021530672a908a@xxxxxxxxxxxxxx> <1104764660.1048.578.camel@xxxxxxxxxxxxxxxx>
Sender: netdev-bounce@xxxxxxxxxxx
User-agent: Gnus/5.1006 (Gnus v5.10.6) XEmacs/21.4 (Corporate Culture, linux)
    jamal> Andi or anyone else - has anyone really done perfomance
    jamal> analysis of LLTX (with anything other than loopback) and
    jamal> shown it has any value? Maybe time to just shoot the damn
    jamal> thing.

For the IPoIB driver, it seems to be worth a couple percent.  It's
hard to get really consistent results but on my pair of test systems,
TCP throughput with netperf goes up by about 30 Mbps out of roughly
1900 Mbps total.  It's probably not worth the complexity in the net
core if all drivers only get that level of improvement.

 - Roland

<Prev in Thread] Current Thread [Next in Thread>