Producer/Consumer Threads

Back to the Win32 Shootout Back to dada's perl lab

N is the number of items produced/consumed.

Measurement of CPU as N varies   N Test Source 10000 30000 70000 100000 bigforth 0.02 0.03 0.04 0.06 smlnj 0.08 0.17 0.35 0.49 erlang 0.49 0.53 0.62 0.68 gnat 0.19 0.52 1.20 1.69 oz 0.31 0.70 1.49 2.14 java 0.69 1.08 1.83 2.40 gcc 0.28 0.76 1.73 2.46 ici 0.31 0.88 2.02 2.86 pike 1.20 3.36 7.71 10.99 csharp 2.70 5.90 12.15 16.85 python 2.75 7.62 18.34 24.76 ruby 12.33 36.13 85.07 120.27 cygperl F F F F perl F F F F ocamlb F F F F mingw32 F F F F pliant T T T T ghc F F F F ocaml F F F F gforth F F F F poplisp T T T T guile F F F F

Measurement of Memory as N varies   N Test Source 10000 30000 70000 100000 bigforth 928 928 928 928 smlnj 936 936 936 936 erlang 5240 5236 5236 5240 gnat 828 828 828 828 oz 648 648 648 648 java 4616 4616 4616 4616 gcc 1512 1512 1512 1512 ici 1144 1144 1144 1144 pike 3592 3592 3600 3600 csharp 3692 3776 3776 3772 python 2620 2620 2620 2620 ruby 2060 2068 2060 2060 cygperl F F F F perl F F F F ocamlb F F F F mingw32 F F F F pliant T T T T ghc F F F F ocaml F F F F gforth F F F F poplisp T T T T guile F F F F

The following graph is similar to the graph on the index page, but here, N is much smaller, and startup costs may be more prominent.

[Note: Values have been normalized to fall in the range of 0-10 for aesthetic reasons. Original value ranges are included on the X-axis.

Source Code CPU (sec) Mem (KB) Lines Code Log bigforth 0.02 928 25 log smlnj 0.08 936 178 log gnat 0.19 828 0 log gcc 0.28 1512 70 log oz 0.31 648 86 log ici 0.31 1144 42 log erlang 0.49 5240 32 log java 0.69 4616 76 log pike 1.20 3592 34 log csharp 2.70 3692 53 log python 2.75 2620 40 log ruby 12.33 2060 34 log poplisp T T 40 log ocamlb F F 31 log mingw32 F F 70 log guile F F 35 log perl F F 38 log cygperl F F 38 log gforth F F 26 log pliant T T 55 log ocaml F F 31 log ghc F F 38 log