o Raphael Coeffic [06/04/08 12:48]: > Stefan Sayer wrote: >> Hi Raphael, >> >> even though the comment had been wrong before (2 us, not ms), I am not >> sure whether I would change this without testing the accuracy of the >> timer in high load situations. With the current value I have seen >> quite good accuracy, what were your results with the value of 2ms? >> >> > The point is that there is no need to go sleeping for 2 ms.... Most of > the time, calling nanosleep with such a value will end up in a very > expensive wait (under some circumstances, even an active wait based on > CPU loop...). I don't think that is true any more for newer linux kernels, and most definitely not for high res timers kernel, which anyone running multimedia apps will probably have enabled.
Stefan > > If you are unsure, you can of course, test this extensively, but i am > pretty sure that it is safe to correct this. > > -Raphael. >> Stefan >> >> o [EMAIL PROTECTED] [06/04/08 10:21]: >> >>> Author: rco >>> Date: 2008-06-04 10:21:07 +0200 (Wed, 04 Jun 2008) >>> New Revision: 1004 >>> >>> Modified: >>> branches/1.0.0/core/AmMediaProcessor.cpp >>> Log: >>> - fixed minor bug. (trunk 1002:1003) >>> >>> >>> Modified: branches/1.0.0/core/AmMediaProcessor.cpp >>> =================================================================== >>> --- branches/1.0.0/core/AmMediaProcessor.cpp 2008-06-04 08:19:31 >>> UTC (rev 1003) >>> +++ branches/1.0.0/core/AmMediaProcessor.cpp 2008-06-04 08:21:07 >>> UTC (rev 1004) >>> @@ -204,7 +204,7 @@ >>> sdiff.tv_sec = diff.tv_sec; >>> sdiff.tv_nsec = diff.tv_usec * 1000; >>> >>> - if(sdiff.tv_nsec > 2000) // 2 ms >>> + if(sdiff.tv_nsec > 2000000) // 2 ms >>> nanosleep(&sdiff,&rem); >>> } >>> >>> >>> _______________________________________________ >>> Semsdev mailing list >>> [email protected] >>> http://lists.iptel.org/mailman/listinfo/semsdev >>> >> >> > -- Stefan Sayer VoIP Services [EMAIL PROTECTED] www.iptego.com iptego GmbH Am Borsigturm 40 13507 Berlin Germany Amtsgericht Charlottenburg, HRB 101010 Geschaeftsfuehrer: Alexander Hoffmann _______________________________________________ Semsdev mailing list [email protected] http://lists.iptel.org/mailman/listinfo/semsdev
