Tamarin Benchmarks Part 3

Here's another benchmark of a simple for loop. The benchmark used is:

var j = 0;
for (var i = 0; i < 100000000; i++) {
     j++;
}
~        

And here are the results:

Tamarin-Tracing Single Interpreter - Compiler on
time ./avmplus ~/Projects/tamarin-central/test/custom/forLoop.abc
real    0m13.121s
user    0m13.064s
sys    0m0.056s
Tamarin-Tracing Single Interpreter - Compiler off
time ./avmplus -interp ~/Projects/tamarin-central/test/custom/forLoop.abc
real    1m33.280s
user    1m33.060s
sys    0m0.195s
Tamarin-Tracing Double Interpreter - Compiler on
time ./avmplus ~/Projects/tamarin-central/test/custom/forLoop.abc
real    0m12.467s
user    0m12.441s
sys    0m0.025s
Tamarin Tracing Double Interpreter - Compiler off
time ./avmplus -interp ~/Projects/tamarin-central/test/custom/forLoop.abc
real    3m41.578s
user    3m41.006s
sys    0m0.412s
Tamarin-Central - Compiler
time ./shell -Dforcemir ~/Projects/tamarin-central/test/custom/forLoop.abc
real 0m6.232s
user 0m6.224s
sys 0m0.007s
Tamarin-Central - Interpreter
time ./shell -Dinterp ~/Projects/tamarin-central/test/custom/forLoop.abc
real    0m36.891s
user    0m36.796s
sys    0m0.094s
SpiderMonkey 1.7
time js forLoop.js
real    0m6.789s
user    0m6.776s
sys    0m0.013s
SpiderMonkey 1.8
time ./js ~/Projects/tamarin-central/test/custom/forLoop.js
real    0m6.396s
user    0m6.385s
sys    0m0.011s

This doesn't bode too well for Tamarin....At least the new Interpreter is in general faster. Again, these benchmarks are not really comprehensive, but it is simply insight as to how Tamarin is shaping up against SpiderMonkey.

* Edit: forced tamarin-central to compile everything. Results updated.

Comments

[Andreas Gal]

Could you please re-run the same benchmark, but this time type-annotate the loop counter as int.

[Brendan Eich]

It would be interesting for such a simple benchmark to see what instructions Tamarin-Tracing records and optimizes.

/be