[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <1366403483.16391.59.camel@edumazet-glaptop>
Date: Fri, 19 Apr 2013 13:31:23 -0700
From: Eric Dumazet <eric.dumazet@...il.com>
To: Stephen Hemminger <stephen@...workplumber.org>
Cc: David Miller <davem@...emloft.net>, gregkh@...uxfoundation.org,
netdev@...r.kernel.org
Subject: Re: [RFC 1/2] netmap: infrastructure (in staging)
On Fri, 2013-04-19 at 13:16 -0700, Stephen Hemminger wrote:
> I can not get line rate output with pktgen on existing kernels today.
I have no trouble saturating at line rate with pktgen, and using
multiqueue NIC.
> I can get line rate easily with netmap.
Yes, but then, its about bypassing the OS, and reimplementing everything
in user land. TCP/UDP/IP stack, iptables, rate limiting, sharing a NIC
among users of this NIC, adding new hot points (sending tx descriptors
without caring about need_resched or whatever)
But I agree pktgen is kind of limited (it sends UDP packets.), please do
not compare pktgen and netmap.
pktgen was something to avoid spending time in UDP/IP stack, and it was
probably needed years ago or if network researchers want to use a laptop
as a pktgen host.
It reminds me the days linux had http server in kernel.
With 8+ cpus, you can plainly use a user land application.
I really hope we do not use pktgen as an argument for having netmap in
the kernel.
--
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