[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux1394-devel
Subject:    Re: [PATCH] Use posted writes for firewire-net.
From:       Stefan Richter <stefanr () s5r6 ! in-berlin ! de>
Date:       2011-10-10 20:08:50
Message-ID: 20111010220850.37835642 () stein
[Download RAW message or body]

On Oct 10 Stephan Gatzka wrote:
> It's interesting that the performance only increased by these few per 
> cents on your systems.
> 
> On our embedded systems (ppc MPC5200 400 Mhz and ARM Cortex A8 with 1.5 
> GHz) we had an increase between 10% and 30% but we did all measurements 
> with iperf.

One factor is whether the response packets cause additional arbitration
gaps, or don't.  I'm not sure about that and too lazy to look it up right
now. :-)  Another aspect is whether the generation of response packets
causes significant CPU overhead.  The packet rate is in the order of
magnitude of 13 kHz ( = 20 MB/s / 1500 B/packet), but how many cycles per
packet might be spent?
-- 
Stefan Richter
-=====-==-== =-=- -=-=-
http://arcgraph.de/sr/

------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure contains a
definitive record of customers, application performance, security
threats, fraudulent activity and more. Splunk takes this data and makes
sense of it. Business sense. IT sense. Common sense.
http://p.sf.net/sfu/splunk-d2dcopy1
_______________________________________________
mailing list linux1394-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux1394-devel
[prev in list] [next in list] [prev in thread] [next in thread]