darcs patch: Teach the inliner about primitive compar... (and 2more)

[Samuel Bronson]

On 10/17/06, Simon Peyton-Jones <simonpj at microsoft.com> wrote:
> We'll look again at those profiling bugs -- but meanwhile there's a
> difficulty lurking: profiling will change what the optimiser does, which

That's not nice! I mean, of course it will affect the speed and
*actual* memory usage... but it would be nice if it could get the
optimizer to just play along, so you could get decent allocation
profiles...

> may conceal exactly what you are looking for.  Instead we usually use

Indeed, it seems to. At least, if I have cost centers on the
sub-module level, it does...

> valgrind to profile execuction at a lower level http://valgrind.org.
> Something like this
>
>         valgrind --tool=cachegrind <prog> <args>
>         [ creates a file like cachegrind.out.23915]
>         cg_annotate --23915
>         [ use the correct number, of course ]

Does that help track down allocation? If so, how?