site_archiver@lists.apple.com Delivered-To: Darwin-dev@lists.apple.com Le 16 déc. 05 à 21:57, Markus Hitter a écrit : Am 16.12.2005 um 15:06 schrieb Stéphane Letz: Hum... It seems you've not really read/understood my mail... Regards Stephane Letz _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.appl... We are debugging a strange issue with abnormal CPU spikes occurring in a test application. CPU spikes are perfectly normal. If you start some calculation, the CPU runs 100%, until done. If you do a short calculation, you get a spike. Anything else would be a waste of time without a benefit. The question you're after is probably, how are other tasks/apps/ threads affected by your app. This is where scheduling policies jump into the picture. The thread that show this behaviour is a real-time (time-constraint) priority 97, and in the test applications (the *simplest* code i had to show the problem) does nothing more that rescheduling itseld the next millisecond. Thus it normal bahaviour is to consume almost 0% The real application does *something* in its real-time thread, but its timing get disturbed when the CPU spike (which last several second) occurs. This email sent to site_archiver@lists.apple.com