On Thu, Sep 26, 2002 at 09:55:46PM +0900, Bang Jun-Young wrote:
> > 
> > If I were to hazard a guess, I'd say you have the causality wrong
> > here.  The fact that static executable startup is slower is likely a
> > kernel issue that will also affect (possibly even worse) dynamic
> > executables.
> 
> Maybe due to lack of memcpy/copyin optimized for recent x86 processors
> in our kernel?

Well, run a profiling kernel and see what's up. I noticed that execing
ELF binaries was quite a bit slower at the time we switched to them,
but I didn't do any detailed measurements. pmc(1) seemed to indicate
less optimal memory behavior (lots of TLB misses), but my memory
isn't clear on that anymore.

- Frank

-- 
Frank van der Linden                                    fvdl@wasabisystems.com
==============================================================================
Quality NetBSD Development, Support & Service.   http://www.wasabisystems.com/