monitoring cache performance of user programs?

[David Young <dyoung%pobox.com@localhost>]

I'm wondering if anybody is using pmc(1) to monitor cache performance of
user programs on modern x86?  Maybe there is a better way to do this in
NetBSD, today, such as a facility provided by DTrace?

Ultimately I would like to have a performance profile of some of my
programs that tells not only how long is spent in each line, but roughly
how the time was spent, so that I can make better decisions about
whether/how to optimize the code.  Even if time is ascribed to causes
(cache misses, page faults) in a very approximate way, I think that it
would be more illuminating than simulating the code in my head.

Dave

-- 
David Young
dyoung%pobox.com@localhost    Urbana, IL    (217) 721-9981