netdev
[Top] [All Lists]
<prev [Date] next>
[Advanced]
 <prev [Thread] next>

Re: 3c59x (was Route cache performance under stress)

from [Andi Kleen] [Permanent Link][Original]
To: Bogdan Costescu <bogdan.costescu@xxxxxxxxxxxxxxxxxxxxx>
Subject: Re: 3c59x (was Route cache performance under stress)
From: Andi Kleen <ak@xxxxxxx>
Date: Tue, 10 Jun 2003 18:49:49 +0200
Cc: "David S. Miller" <davem@xxxxxxxxxx>, sim@xxxxxxxxxxxxx, ralph+d@xxxxxxxxx, hadi@xxxxxxxxxxxxxxxx, xerox@xxxxxxxxxx, fw@xxxxxxxxxxxxx, netdev@xxxxxxxxxxx, linux-net@xxxxxxxxxxxxxxx
In-reply-to: <Pine.LNX.4.44.0306101815550.26879-100000@xxxxxxxxxxxxxxxxxxxxxxxxxxx>
References: <20030610.085600.71109220.davem@xxxxxxxxxx> <Pine.LNX.4.44.0306101815550.26879-100000@xxxxxxxxxxxxxxxxxxxxxxxxxxx>
Sender: netdev-bounce@xxxxxxxxxxx 
> 
> And another important question: how much improvement can be gained from
> the driver ? Folks that do parallel computation over TCP over Ethernet

You can play some tricks with the driver to make eth_type_trans disappear
from the profiles. This usually helps a lot because it avoids one
full "fetch from cache cold memory" roundtrip per packet, which is slow on
any CPU.

-Andi
[More with this subject...]
<Prev in Thread] Current Thread [Next in Thread>
Previous by Date:  Re: 3c59x (was Route cache performance under stress), Bogdan Costescu
Next by Date:  Re: [patch]: CONFIG_IPV6_SUBTREES fix for MIPv6, David S. Miller
Previous by Thread:  Re: 3c59x (was Route cache performance under stress), Bogdan Costescu
Next by Thread:  Re: 3c59x (was Route cache performance under stress), Robert Olsson
Indexes:  [Date] [Thread] [Top] [All Lists]