[MLton-user] timing anomoly

[Sean McLaughlin]

Hi,

  Using -inline 1000 worked for the test example, but in the context of a large
program it is now taking a very long time to compile.  I killed the
process after
30 minutes.  Is there a way to get the fast behavior (%10 faster than GCC -O2)
without such compilation time increases?  I know the comparison is not
really relevant, but
GCC manages to generate the fast code in a few seconds, even in a program
of the same order of magnitude of my SML program.

Best,

Sean

On Dec 3, 2007 6:48 PM, Sean McLaughlin <seanmcl at gmail.com> wrote:
> Thanks.
>
> It worked with -inline 1000.  Will this negatively impact performance
> if I use this option on the rest of my code?  Also, is there documentaton for
> the -expert options?  I don't want to waste your time if there is a document
> somewhere describing these options.
>
> Best,
>
> Sean
>
>
> On Dec 3, 2007 5:05 PM, Florian Weimer <fw at deneb.enyo.de> wrote:
> > * Sean McLaughlin:
> >
> > >   I found some very strange behavior of mlton while running some
> > > floating point experiments.  The attached file evaluates some
> > > polynomials and returns them.  If you multiply a value by a single
> > > argument, it multiplies by 10 the performance time of the entire
> > > function.
> >
> > It seems as if the inline limit exceeded after the change.  Try
> > something like "-inline 500" and see if it improves performance.
> >
>