Comments on: AS3 vs. JavaScript Performance Followup (March 2010)

By: Mario Klingemann

Mario Klingemann — Wed, 10 Mar 2010 17:48:05 +0000

Very interesting results. Good that I did not bet any money on that I would have sworn that Vectors make it faster.

By: jackson

jackson — Thu, 04 Mar 2010 19:33:28 +0000

Yes, that is the test I'm using. You can find a link in the spreadsheets and the original test, but I stopped adding it to the followup articles. I converted the test to use Vector instead of Array just now and found the following results:

Test 0: no change
Test 1: massively slower (2-3x) due to the required sort function. See my article about this.
Test 2: no change
Test 3: no change
Test 4: no change
Test 5: no change
Test 6: no change
Test 7: no change
Test 8: no change
Test 9: no change
Test 10: no change
Test 11: no change
Test 12: no change

So Vector doesn't seem to help AS3/Flash Player's performance overall.

By: jackson

jackson — Thu, 04 Mar 2010 18:55:45 +0000

Tomorrow’s article will be on the subject of different increment and decrement strategies.

By: Rezmason

Rezmason — Thu, 04 Mar 2010 14:18:00 +0000

Yes, please make sure you use typed AS3 Vectors instead of Arrays in the performance test. Even if a SWF does not contain user-written ActionScript, Vectors are still used internally by the player.

Also, for what it’s worth I hear (i += 1) is faster than (i++) in ActionScript.

By: Mario Klingemann

Mario Klingemann — Thu, 04 Mar 2010 11:58:37 +0000

That’s some very interesting info – thanks! But I had a look at the ActionScript test code (I assume you use the one by Ted Patrick here: http://www.onflex.org/perf/srcview/ ) and it seems like all the tests are still using Arrays. In a real world situation AS3 developers would use the new Vector class for these kinds of operations so I think this test does not correctly reflect the possible speed AS3 could achieve.