I have no insights without profiling the code paths which I have done for
iperf2 but not for iperf3.
Bob
From: Jon Lederman [mailto:jonleder...@gmail.com]
Sent: Tuesday, September 02, 2014 8:47 PM
To: Bob (Robert) McMahon
Cc: iperf-users@lists.sourceforge.net; George Halpert
Subject: Re: [Iperf-users] Difference In Jitter Calculations Between IPerf2 and
IPerf3
Thanks for sending this. Do you have any insights into the reasons for these
differences. We will send you our results as well.
On Sep 2, 2014, at 5:20 PM, Bob (Robert) McMahon
<rmcma...@broadcom.com<mailto:rmcma...@broadcom.com>> wrote:
Ok, I see different values as well. Iperf2 is showing approximately 20 us
jitter while iperf3 is showing ~150 us. This is using a 200Mbs udp stream on
802.11 AC.
Bob
From: Bob (Robert) McMahon [mailto:rmcma...@broadcom.com]
Sent: Tuesday, September 02, 2014 1:34 PM
To: Jon Lederman
Cc: iperf-users@lists.sourceforge.net<mailto:iperf-users@lists.sourceforge.net>
Subject: Re: [Iperf-users] Difference In Jitter Calculations Between IPerf2 and
IPerf3
Don't know but comments are ignored by the compiler ;)
Can you share your results? I'd be curious to see the variation and if it
could be related to things like scheduling.
Also, what os are you using? If you are using linux you may want to set the
processes to use SCHED_RR
http://www.cyberciti.biz/faq/howto-set-real-time-scheduling-priority-process/
I do that in iperf2
http://sourceforge.net/projects/iperf2/?source=directory
Bob
From: Jon Lederman [mailto:jonleder...@gmail.com]
Sent: Tuesday, September 02, 2014 1:29 PM
To: Bob (Robert) McMahon
Cc: iperf-users@lists.sourceforge.net<mailto:iperf-users@lists.sourceforge.net>
Subject: Re: [Iperf-users] Difference In Jitter Calculations Between IPerf2 and
IPerf3
Thanks. Look forward to your results. Even though I also noticed that the
calculation looks similar on the surface, I did notice a comment in iperf3 to
the effect:
// XXX: This is NOT the way to calculate jitter
// J = |(R1 - S1) - (R0 - S0)| [/ number of packets, for average]
Any thoughts why that is in there? The calculation in the code appears to have
come directly out of the RFC 1889/3550.
Also, I want to clarify that this calculation seems to be average jitter not
instantaneous jitter.
Thanks.
-Jon
On Sep 2, 2014, at 4:23 PM, Bob (Robert) McMahon
<rmcma...@broadcom.com<mailto:rmcma...@broadcom.com>> wrote:
A quick cursory view of the source suggests the same calculation. Let me run
both on my test rig and see if I get similar results.
Bob
-----Original Message-----
From: Jon Lederman [mailto:jonleder...@gmail.com]
Sent: Monday, September 01, 2014 5:22 PM
To: iperf-users@lists.sourceforge.net<mailto:iperf-users@lists.sourceforge.net>
Subject: [Iperf-users] Difference In Jitter Calculations Between IPerf2 and
IPerf3
Hi,
In measuring jitter using Iperf 2.0.5 and Iperf 3, I have noticed a marked
difference in the calculations up to an order of magnitude. Could you explain
the differences in calculation of jitter between 2.0.5 and 3 and the accuracy.
Which should I rely upon?
Thanks in advance.
Jon
------------------------------------------------------------------------------
Slashdot TV.
Video for Nerds. Stuff that matters.
http://tv.slashdot.org/
_______________________________________________
Iperf-users mailing list
Iperf-users@lists.sourceforge.net<mailto:Iperf-users@lists.sourceforge.net>
https://lists.sourceforge.net/lists/listinfo/iperf-users
------------------------------------------------------------------------------
Slashdot TV.
Video for Nerds. Stuff that matters.
http://tv.slashdot.org/
_______________________________________________
Iperf-users mailing list
Iperf-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/iperf-users
