Re: iperf buffer again.

To: chongz --at-- positioning-research.com
Subject: Re: iperf buffer again.
From: Kevin Gibbs <kgibbs --at-- ncsa.uiuc.edu>
Date: Fri, 6 Dec 2002 13:29:46 -0600 (CST)
Cc: iperf-users --at-- dast.nlanr.net
Content-type: TEXT/PLAIN; charset=US-ASCII
In-reply-to: <495487839.1039201599706.JavaMail.root --at-- orion>
Reply-to: Kevin Gibbs <kgibbs --at-- ncsa.uiuc.edu>
Sender: owner-iperf-users --at-- dast.nlanr.net

On Fri, 6 Dec 2002 chongz --at-- positioning-research.com wrote:

> PS: One quick question though: if ws=2, how could I get 128K -> 95568?

I did not say that WS always equals 2, I ran some tests with my machine 
and with the buffer size of 128K the connection used WS=1 and for 240K it 
used WS=2, therefore I used those guesses to decipher your results so 
47784 * 2 equals 95568 it surely wouldn't be 47784 * 4 since that is 
143436 and larger than 128K, though that could happen with some OSes, it 
would not be the case for the version of Linux you are running.

Kevin

References:
- Re: iperf buffer again.
  - From: chongz --at-- positioning-research.com

Prev by Date: Some questions about Iperf
Next by Date: Re: Iperf and latency
Previous by thread: Re: iperf buffer again.
Next by thread: Iperf and latency

Other Mailing lists | Author Index | Date Index | Subject Index | Thread Index