contification paper: benchmarks

Matthew Fluet fluet@CS.Cornell.EDU
Sun, 11 Mar 2001 22:38:32 -0500 (EST)


> The compilations of the kit have finished.  The log follows.  I've put the
> executables at http://www.star-lab.com/sweeks/kit.tgz.  They should be run with
> the in a directory with DATA/basislib/ available.  I'll leave the mlton
> compilations running overnight, and send the results in the morning.

> mlton -contify dom
> functions: 7381
>   call_cont_dom: 2216  call_cont: 0
>   call_dom: 560  cont_dom: 354
>   call: 0  cont: 0  dom: 272
>   rem_call_cont_dom: 3  rem_call_cont: 0
>   rem_call_dom: 0  rem_cont_dom: 0
>   rem_call: 0  rem_cont: 0  rem_dom: 0
> nested: 21  rejected: 3

Great.  We're now up to > 6200 contified functions, with only 59 nested
and 25 rejected.  A self compile has no rejections, but I don't remember
about the nesting.  But it's also got lots of contified functions.