Home My Page Projects Code Snippets Project Openings SML/NJ
Summary Activity Forums Tracker Lists Tasks Docs Surveys News SCM Files

SCM Repository

[smlnj] View of /sml/trunk/benchmarks/programs/nucleic/MAIL
ViewVC logotype

View of /sml/trunk/benchmarks/programs/nucleic/MAIL

Parent Directory Parent Directory | Revision Log Revision Log


Revision 193 - (download) (annotate)
Fri Nov 20 17:43:59 1998 UTC (22 years, 10 months ago) by monnier
File size: 1184 byte(s)
Initial revision
Here is an sml program translated from scheme by Marc Feeley.  It does
some calculations relating to the 3D structure of nucleic acids.
Since it exists now in C, scheme, sml, miranda, and haskell versions
people here were doing some benchmarking.  The C version takes about 4
seconds on a sparc SLC machine (which I think is around 20 integer
specmarks).  They had sml/nj 0.75 installed here, and this sml version
ran in 11 seconds on the SLC (after taking _lots_ of time and memory
to compile).  I thought it might be faster on a new version, so I
brought 1.03f over and built it.  Surprisingly, 1.03f ran it in 22 seconds.

The translation is crude, so it could probably be speeded up considerably.
But why the 100% slowdown between 0.75 and 1.03f?  Is it the space-unsafe
optimizations in 0.75 again?  It would be good to profile this and see
what's happening.  I haven't had time to look at the code, but it has
some big tuples of tuples of reals, which probably explains the slow
compilation.

By the way, the final answer should be 179.

Dave

P.S. The Glasgow Haskell compiler never finished compiling the Haskell
version, and Harlequin MLWorks dumped core while compiling it.

root@smlnj-gforge.cs.uchicago.edu
ViewVC Help
Powered by ViewVC 1.0.0