netdev
[Top] [All Lists]

Re: More measurements

To: kuznet@xxxxxxxxxxxxx
Subject: Re: More measurements
From: Andrew Morton <andrewm@xxxxxxxxxx>
Date: Wed, 31 Jan 2001 11:30:48 +1100
Cc: netdev@xxxxxxxxxxx
References: <3A76C5DE.993BB140@xxxxxxxxxx> from "Andrew Morton" at Jan 30, 1 04:45:00 pm <200101301914.WAA12591@xxxxxxxxxxxxx>
Sender: owner-netdev@xxxxxxxxxxx
kuznet@xxxxxxxxxxxxx wrote:
> 
> Hello!
> 
> > ** Yup, I just retested everything.  With vanilla 2.4.1,
> > when data is sent with sendfile() we get 25% more packets
> > coming back than when sending with 64 kbyte send()s.  And
> > 25% more bytes.
> 
> This was due to pushes made by 4K writes used by old sendfile().
> 
> Pretty silly feature, but it seems to be required. 25% of packets
> are better than full collapse in some cases. Sigh...

Ah.  Thanks.  It's good to have an explanation :)

> In any case, I did not understand one thing:
> do you really see some case, when you do not saturate eepro100?
> This is sort of difficult to make.

In my testing, in all cases, at all times, with all NICs, the
link is 100% saturated.  I don't need another dimension to
deal with!

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