Another look at Tamarin Benchmarks

Given the results from the previous benchmark post, I figured I would try JGFsor as recommended by Andreas. I tweaked the source a bit to do work on a smaller data set as the original data set took too long for the Tamarin interpreters. Here are the results:

Tamarin-Tracing Single Interpreter - Compiler on
time ./avmplus ~/Projects/tamarin-central/test/custom/JGFsor.abc

real 0m2.900s
user 0m2.736s
sys 0m0.039s

Tamarin-Tracing Single Interpreter - Interpreter on
time ./avmplus -interp ~/Projects/tamarin-central/test/custom/JGFsor.abc

real 1m0.979s
user 1m0.895s
sys 0m0.025s

Tamarin-Tracing Double Interpreter - Compiler on
time ./avmplus ~/Projects/tamarin-central/test/custom/JGFsor.abc

real 0m2.894s
user 0m2.864s
sys 0m0.026s

Tamarin-Tracing Double Interpreter - Interpreter on
time ./avmplus -interp ~/Projects/tamarin-central/test/custom/JGFsor.abc

real 0m50.084s
user 0m50.025s
sys 0m0.028s

Tamarin-Central Interpreter
time ./shell -Dinterp ~/Projects/tamarin-central/test/custom/JGFsor.abc

real 0m55.597s
user 0m55.360s
sys 0m0.167s

Tamrin-Central Compiler
time ./shell ~/Projects/tamarin-central/test/custom/JGFsor.abc

real 0m19.715s
user 0m19.543s
sys 0m0.110s

Spidermonkey 1.7
time ./js ~/Projects/tamarin-central/test/custom/JGFsor.js

real 0m8.267s
user 0m8.148s
sys 0m0.096s

Spidermonkey 1.8
time ./js ~/Projects/tamarin-central/test/custom/JGFsor.js
real 0m5.170s
user 0m4.576s
sys 0m0.589s

At least Tamarin beat SpiderMonkey this time. However, I believe the issue of the speed difference is still in the array access. I'm going to do one more quick benchmark which will be posted shortly.