Hi Konrad, why should it put load on all cores if there's not enough workload to saturate even one?
When you have one thread running at 100% for several seconds, that thread will also stay on the same core because there's overhead involved in switching cores (core-local cache warmup). So if there aren't really several requests that actually run concurrently, why should the scheduler cycle through the cores? Maik Am 25.04.2013 um 14:36 schrieb Konrad Kubacki <[email protected]>: > Hi, > > I've set WOAllowsConcurrentRequestHandling to true and thought that with this > option enabled, WebObjects gonna consume all of my 4 CPU cores. > I checked production server statistics, which shows as: (see the picture: > http://d.pr/i/qJoH) Yes, i know - the picture shows the load of all processes > - which my not work in multicore env, but still, there are 2x more threads of > WOApp on core 0. > > See threads for one of the application's instances are as follow: > CORE : NUM_THREADS > 0 : 41 > 1 : 12 > 2 : 13 > 3 : 14 > > (done by ps -p PID -L -o psr | sort | uniq -c) > > Why it utilizes much more core0? Is there a way to enable better consumptions > of all the cores? (all but core 0 looks quite lazy at the time) > > Konrad > _______________________________________________ > Do not post admin requests to the list. They will be ignored. > Webobjects-dev mailing list ([email protected]) > Help/Unsubscribe/Update your Subscription: > https://lists.apple.com/mailman/options/webobjects-dev/maik%40selbstdenker.ag > > This email sent to [email protected] _______________________________________________ Do not post admin requests to the list. They will be ignored. Webobjects-dev mailing list ([email protected]) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/webobjects-dev/archive%40mail-archive.com This email sent to [email protected]
