See this discussion on kerneltrap.org: http://kerneltrap.org/node/view/3429 Quoting Con Kolivas: "Last time I posted this patch to lkml [the Linux kernel mailing list] almost noone responded on list to say it helped their workload, which is why it was never taken seriously by akpm. Strangely, once again, noone seems to have anything good to say about it in the formal confines of lkml so the list (and myself) will believe it is not worth pursuing." I think Con is right. I think the core Linux kernel developers are really not aware that the vanilla kernel is not appropriate for multimedia work. And that's because no one told them. As a result, i started a thread on the linux-kernel mailing list (named "desktop and multimedia as an afterthought?") in which i try to describe the issues with the Linux kernel on multimedia systems. http://marc.theaimsgroup.com/?t=108965898400007&r=1&w=2 Please subscribe to LKML and reply to the thread. You can subscribe here: http://vger.kernel.org/majordomo-info.html It looks like the kernel people want "numbers". I.e., to quote Denis Vlasenko: "Hi. I tested patches X,Y and Z. X have latency spikes of NN ms. Y does not have 'em, but kernel compile time have gone up by MM mins. Z does not have latency spikes _and_ do not have performance regression like Y. I am willling to test more patches." But i suspect any kind of informative and objective report would be fine. Thank you. -- Florin Andrei http://florin.myip.org/
