Re: [PATCH] improving O(1)-J9 in heavily threaded situations

[Ingo Molnar]

On Tue, 5 Feb 2002, Jussi Laako wrote:

> Ok, now I made comparison test with exactly same kernel except other
> one with -K2 patch. O1-K2 behaves significantly worse than old
> scheduler. I think this behaviour was introduced somewhere around
> beginning of -J series. I can't make kernel with old scheduler loose
> datablocks, but with O1 it looses large percentage of the blocks.

what does 'loose datablocks' mean? What application loses datablocks?

	Ingo