>>>>> On Tue, 21 Sep 2004 15:35:05 -0700, Ben Collver <ben@NetBSD.org> said:

> ``And there are also some cases where we took a direction with the
> underlying technology that turned out to be a mistake. An example is the
> two-level thread scheduling model, where thread scheduling happens both
> at user level and in the kernel.
	:
> it turned out to be not only much simpler and easier to maintain,
> but also faster in almost every case.

It may be true that Solaris M:N thread is slower than Solaris 1:1
thread, but that doesn't mean NetBSD M:N thread is worse than 1:1
threads.

Acutally I did some benchmarks about basic therad primitives
last year. And NetBSD SA is faster than all of the followings:
	- FreeBSD 5 threads (M:N)
	- NPTL - new Linux threads (1:1)
	- old Linux threads (1:1)

I suspect the reason why Solaris M:N thread is slower is because
generally it doesn't assign enough lwps to pthreads. NetBSD SA
doesn't have the problem.
--
soda