[MLton-devel] benchmarks on RedHat and FreeBSD
Stephen Weeks
MLton@mlton.org
Thu, 19 Sep 2002 12:21:23 -0700
Here are the results for all our usual benchmarks, run on both RedHat
7.1 and FreeBSD 4.6.2 vmware machines with 500M RAM, 1.6Ghz.
It looks like FreeBSD is often a lot slower. From looking at a couple
of runs, I conjecture much of it is due to the absence of mremap and
zeroing costs on FreeBSD. For example, using a fixed heap of 100m for
mlyacc cuts the ratio down from 2.5 to 1.6. I don't know what the
rest of the slowdown could be due to. I suppose for nucleic and zern
it could be double alignment.
MLton0 -- mlton on RedHat 7.1
MLton1 -- mlton on FreeBSD 4.6.2
run time ratio
benchmark MLton1
barnes-hut 1.1
boyer 1.7
checksum 1.1
count-graphs 1.2
DLXSimulator 1.8
fft 1.2
fib 1.1
hamlet 3.0
imp-for 1.1
knuth-bendix 1.5
lexgen 1.2
life 1.3
logic 1.0
mandelbrot 1.0
matrix-multiply 1.2
md5 1.1
merge 1.1
mlyacc 2.5
mpuz 1.0
nucleic 3.0
peek 1.2
psdes-random 1.1
ratio-regions 1.4
ray 1.2
raytrace 1.2
simple 1.5
smith-normal-form 1.8
tailfib 1.0
tak 1.0
tensor 1.0
tsp 1.0
tyan 1.4
vector-concat 1.8
vector-rev 1.1
vliw 1.3
wc-input1 1.5
wc-scanStream 1.7
zebra 1.2
zern 2.7
compile time
benchmark MLton0 MLton1
barnes-hut 2.1 2.9
boyer 4.4 7.9
checksum 0.6 1.0
count-graphs 1.4 2.0
DLXSimulator 3.7 7.4
fft 1.2 2.0
fib 0.5 0.9
hamlet 43.1 66.1
imp-for 0.6 1.0
knuth-bendix 1.7 2.7
lexgen 4.3 8.1
life 1.1 1.9
logic 2.4 4.7
mandelbrot 0.6 0.9
matrix-multiply 0.6 1.1
md5 1.0 1.6
merge 0.6 1.0
mlyacc 20.8 39.5
mpuz 0.8 1.4
nucleic 4.2 5.2
peek 0.9 1.6
psdes-random 0.6 1.1
ratio-regions 2.3 4.3
ray 2.8 5.1
raytrace 8.1 16.4
simple 5.4 11.5
smith-normal-form 6.0 12.1
tailfib 0.5 0.9
tak 0.5 0.9
tensor 2.5 5.7
tsp 1.3 2.9
tyan 3.1 5.2
vector-concat 0.6 1.1
vector-rev 0.7 1.0
vliw 10.0 13.5
wc-input1 1.4 2.5
wc-scanStream 1.4 2.3
zebra 5.7 10.4
zern 1.2 1.6
run time
benchmark MLton0 MLton1
barnes-hut 2.0 2.1
boyer 3.2 5.5
checksum 1.7 1.9
count-graphs 2.0 2.4
DLXSimulator 14.3 26.4
fft 8.6 10.1
fib 1.5 1.6
hamlet 5.5 16.4
imp-for 3.4 3.8
knuth-bendix 3.9 6.0
lexgen 6.6 8.2
life 4.2 5.6
logic 11.3 11.8
mandelbrot 3.3 3.4
matrix-multiply 1.3 1.5
md5 1.1 1.2
merge 37.9 40.1
mlyacc 8.6 21.8
mpuz 2.5 2.6
nucleic 6.4 19.4
peek 0.4 0.5
psdes-random 1.8 1.9
ratio-regions 5.7 7.9
ray 1.6 2.0
raytrace 2.4 2.8
simple 3.5 5.2
smith-normal-form 0.4 0.7
tailfib 6.5 6.5
tak 4.3 4.4
tensor 1.7 1.7
tsp 3.1 3.1
tyan 10.3 14.8
vector-concat 1.7 3.0
vector-rev 2.7 2.9
vliw 3.8 4.9
wc-input1 1.3 2.0
wc-scanStream 1.5 2.5
zebra 1.3 1.6
zern 23.6 62.6
size
benchmark MLton0 MLton1
barnes-hut 96,225 109,536
boyer 140,649 140,526
checksum 43,641 43,582
count-graphs 64,713 64,526
DLXSimulator 109,713 109,282
fft 52,853 52,754
fib 43,681 43,614
hamlet 1,229,748 1,226,819
imp-for 43,697 43,630
knuth-bendix 86,234 86,173
lexgen 173,489 173,038
life 62,057 61,966
logic 103,697 103,638
mandelbrot 43,777 43,718
matrix-multiply 44,185 44,118
md5 52,882 52,781
merge 44,969 44,910
mlyacc 537,425 536,398
mpuz 47,857 47,790
nucleic 82,729 82,630
peek 51,954 51,853
psdes-random 44,689 44,630
ratio-regions 62,505 62,374
ray 104,644 104,243
raytrace 278,097 277,404
simple 199,829 199,546
smith-normal-form 173,789 187,228
tailfib 43,409 43,342
tak 43,841 43,782
tensor 96,364 109,737
tsp 59,530 59,373
tyan 107,770 107,229
vector-concat 44,185 44,126
vector-rev 44,041 43,974
vliw 325,597 323,866
wc-input1 68,081 68,006
wc-scanStream 68,761 68,694
zebra 144,002 142,373
zern 50,588 50,487
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
MLton-devel mailing list
MLton-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mlton-devel