On Tue, May 06, 2014 at 04:33:38PM +0800, Alex Shi wrote: > > > > > Maybe the predication is reasonable on per task history. but on a cpu > > load history, with many tasks rebalance. No testing show current method > > is helpful. > > > > For task load change, scheduler has no idea for its future except guess > > from its history. but for cpu load change, scheduler know this from task > > wakeup and balance, which both under control and its aim. > > > > > > I think the first patch of this serial has the same effect of LB_LIAS > > disable. and previous result show performance is good. > > > > Anyway, I just pushed the following patch to github, maybe fengguang's > > testing system will care this. > > Fengguang, > > Are there any performance change on > https://github.com/alexshi/power-scheduling.git noload repository?
You forgot to qualify that with the important bit; on _large_ systems. Esp. non fully connected numa boxen. Also, I'm not sure Wu has workloads that are typical of such systems -- even if he has such machines, which I don't know either. Enterprise distro testing has _some_ of that, but the very sad truth is that most enterprise users lag behind at least a full release cycle. So by the time people start using the kernel, its so old nobody really cares anymore :-(
pgpKuK3Q0z2ra.pgp
Description: PGP signature