Time-profile of the DLG event-pipe / sr_waitEvt()

From: "Frode Greger Sivertsen" <frode@netconnect.no>

Probably of little interest, but I thought I'd share this time-profile chart with you as I found it lying around on my disk..

It shows the time it takes to read out N events from the DLG event-pipe with sr_waitEvt(-1).

The bottom line is that applications can benefit from NOT using the Dialogic event-pipe for scheduling user tasks (sr_putEvt). Instead, use IO Completion Ports in order to share a thread-pool between Dialogic & application/user tasks.

Further, it seems that DLG expands the queue as necesarry, but is rather lazy when it comes to freeing up the memory again.

Some optimizations needed, Dialogic guys? 1 minute of 100% CPU (PIII-800) just to read out 30.000 events of 50 bytes each from an array?

 
 - Frode

To reply: mailto:General.26278@pysdiscussext.py.intel.com
To start a new topic: mailto:General@pysdiscussext.py.intel.com
To login: http://members.support.dialogic.com/wbproxy/wbpx.dll/~general
Received on Wed Nov 05 02:42:10 2003