MLton vs OCAML

Stephen Weeks MLton@sourcelight.com
Fri, 6 Oct 2000 13:23:11 -0700 (PDT)


I translated the PLClub ICFP entry from OCAML to SML.  This is a fairly
allocation and double intensive program.  Here are the running times for OCAML
3.00, MLton 20001004, and SML/NJ 110.9.1, as well as some allocation data for
MLton.

		OCAML	MLton SML/NJ		MLton alloc 
		time	time  time		 M	M/s
holes		 1.7	 5.5	12.4		138	25
fov		 1.4	 2.0	11.3		 66	33
intercyl	 1.6	 2.5	13.5		 48	19
snowgoon	 2.9	 4.1	24.4		 33	 8
dice		 3.8	 5.5	29.1		102	19
golf		 1.6	 3.5	11.5		 70	20
cone-fractal	 3.8	 5.1	25.4		 48	 9
large		 4.3	 3.9	13.2		 60	15
pipe		 5.4	 5.5	32.4		114	21
chess		15.9	21.1	99.4		151	 7
fractal		12.1	12.5	86.4		 19	 2

geom-mean	 3.5	 5.1	23.9

As you can see, we still have a ways to go to catch up to OCAML.   

I'm thinking of sending this off to comp.lang.ml and/or sml-implementors.  Any
thoughts or additional info I should provide?