Hello! i am in the middle of planning a large audio application. in order for it to be useful, it needs to have very precise timing control. some simple research has shown that kernel events happen 100 times a second (a jiffy). having a timer time out faster than that results in a timer that is still limited to 1/100th of a second resolution. i need something that is nearer to 1/1000th of a second resolution. the only ways around this that i know of are to possibly use the preemt patch (which requires a kernel upgrade as well), or have the application use 100% cpu time all the time in wasteful loops. recently, pthreads have entered as a potential solution for me. however, i am curious: if you have a task that runs for half a jiffy, then get blocked, does the kernel switch to a new task immediatly, or does it wait for the jiffy to expire before continuing (wasting half a jiffy). i know of no way to test this, so im hoping someone who has worked with the vm might know. thanks! chris ***---cwright@softpixel.com---*** http://softpixel.com/~cwright/ /htgirwc~/moc.lexiptfos//:ptth ***---moc.lexiptfos@thgirwc---*** -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ IRC Channel: irc.openprojects.net / #kernelnewbies Web Page: http://www.kernelnewbies.org/
