On Sun, 06 Jan 2002, martin f krafft wrote: > but any process with nice value -20 should take absolut precedence over > other processes at higher nice levels, especially when the > resource-hog-process runs at nice level +20!!!
But that just will not happen with current unpatched kernels. High IO will starve even RT processes. You will need to patch the kernel with low-latency or preempt patches. You might even need to change the scheduler (yet another patch...). > somehow this strikes me as *wrong*. heck, even windoze NT could do that > better... 2.4.x vanilla is a PoS in certain areas. The VM is one. The latency is another. With dumb xmms, it cannot help but cause skips. Use something better, with huge output buffers, and you will not have so much trouble. Or patch the kernel. -- "One disk to rule them all, One disk to find them. One disk to bring them all and in the darkness grind them. In the Land of Redmond where the shadows lie." -- The Silicon Valley Tarot Henrique Holschuh
