SLED and mlton

[Stephen Weeks]

> I put the mlton source for SLED at
> 
> http://www.eecs.harvard.edu/~nr/sled.sml
> 
> Only 30,000 lines :-)

I compiled this using the following command line

mlton @MLton fixed-heap 350m gc-summary -- -v3 sled.sml

On my .5G machine, it completed in roughly 3.5 minutes, with the following gc
summary info.

max semispace size(bytes): 183,500,800
max stack size(bytes): 2,080,768
GC time(ms): 35,320 (18.0%)
maxPause(ms): 3,170
number of GCs: 43
bytes allocated: 6,621,622,812
bytes copied: 908,077,852
max bytes live: 81,671,552

This is roughly what I'd expect for a 30,000 line program in terms of max bytes
live.  And it shouldn't be causing any problems on a .5G machine.  

We've had reports from another user about strange memory usage problems, and
think it might be due to the kernel reporting the wrong value for the amount of
RAM.  We'd like to track this down.  Do you still see the same "machine to its
knees" problem if you compile with the fixed-heap or max-heap option?