Re: TSO / LSO
Re: TSO / LSO
- Subject: Re: TSO / LSO
- From: Andrew Gallatin <email@hidden>
- Date: Tue, 2 Jan 2007 14:17:31 -0500 (EST)
Ed Wynne writes:
>
> On Jan 2, 2007, at 1:39 PM, Andrew Gallatin wrote:
>
> > The payoff is pretty large. For the 10GbE NIC I do drivers for, we
> > went from sending at ~2Gb/s to ~8Gb/s when using a 1500 byte MTU on a
> > fairly wimpy 1.8GHz single CPU AMD64 when I implemented TSO support
> > in our FreeBSD driver.
>
> Do you have any numbers on the before / after CPU consumption? Doing
> more while using less power appears to be all the rage these days.
> Even better is doing the same thing while using a fraction of the
> power. If through put isn't enough of a motivator, perhaps a little
> extra battery life...
I don't remember the FreeBSD CPU utilization off the top of my head.
I'm pretty sure it was maxed in both cases. But your point is
valid -- the speedup comes mostly from a reduction in CPU usage.
This is more clear under linux, which can actually max the link
with some idle time.
The same machine running linux goes from maxing the link using 32% CPU
to a maxed-out CPU and 5Gbs when TSO is disabled. Eg:
<2:12pm>venice/gallatin:linux>netperf242 -Hrome-my -tTCP_SENDFILE -F /var/tmp/zot -T0,1 -c -C
TCP SENDFILE TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to rome-my (192.168.1.16) port 0 AF_INET : cpu bind
Recv Send Send Utilization Service Demand
Socket Socket Message Elapsed Send Recv Send Recv
Size Size Size Time Throughput local remote local remote
bytes bytes bytes secs. 10^6bits/s % S % S us/KB us/KB
87380 65536 65536 10.01 9430.47 31.99 73.93 0.278 1.284
<2:12pm>venice/gallatin:linux>sudo ethtool -K eth1 tso off
<2:12pm>venice/gallatin:linux>netperf242 -Hrome-my -tTCP_SENDFILE -F /var/tmp/zot -T0,1 -c -C
TCP SENDFILE TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to rome-my (192.168.1.16) port 0 AF_INET : cpu bind
Recv Send Send Utilization Service Demand
Socket Socket Message Elapsed Send Recv Send Recv
Size Size Size Time Throughput local remote local remote
bytes bytes bytes secs. 10^6bits/s % S % S us/KB us/KB
87380 65536 65536 10.00 5008.09 99.90 46.41 1.634 1.518
Drew
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Darwin-kernel mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden