parallel booting init in C - even larger speedup

[Olaf van der Spek]

On 11/14/05, Olivier Sessink <lists at olivier.pk.wau.nl> wrote:
> Olaf van der Spek wrote:
>
> > Isn't an IO bottleneck 'solved' by using readahead?
> > And even without readahead, generating IO requests earlier means the
> > HDD may be able to serve them in a more efficient order.
>
> but you need also more context switches, which increases the overhead.
>
> So the maximum speedup is expected when you boot an IO-bound and a
> CPU-bound process in parallel.

True, but which scripts are completely CPU bound?