On Thu, 11 Nov 2004 14:08:27 +0100
Timo Sch=F6ler <eclipser23@web.de> wrote:

> as i mentioned before, -j n *might* break things up (race conditions),
> so i think it's sensible to go for a -j 1 as default and let the user=20
> device whether (s)he uses -j n (n=B32)... ;)

Yes, I wasn't implying to disregard your post about the race condition,
just a musing about this.

The overall question that I had was whether there is some accounting
that prevents a parent thread from exceeding 100% CPU with its children.

tim