Revised J. Dunstan tests on AS3 vs. JS by Smily

	A	B	C	D	E	F	G	H
1		Rewritten / adjusted tests from http://jacksondunstan.com/articles/2209
2
3		ARTIFICIAL TESTS CAN BE MISLEADING
4		ESPECIALLY THESE ONES D:
5
6		Tested on Windows 7 with a Q6600 at 2.6GHz
7
8		Test	Chrome 26	Firefox 20	IE 9	FP 11.6 (ASC1)	FP 11.6 (ASC2)
9	1	array.join	5	30	30	143	148
10	2	array.sort	450	50	100	145	145
11	3	string concat	2300	1910	6500	36	37
12	4	long concat (too inconsistent)
13	5	harmonic (flops)	150	130	185	115	115
14	6	substring	45	40	65	116	110
15	7	string indexOf	180	330	330	340	345
16	8	Math.round	65	60	3250	180	186
17	9	empty loop	130	130	100	200	175
18	10	func + string + array	60	60	90	163	162
19	11	md5 cupcake	110	120	320	106	104
20	12	func call with return	17	17	130	113	17
21
22
23			x Chrome faster than ASC2	x Firefox faster than ASC2	x IE faster than ASC2		x ASC2 faster than slowest browser	x Fastest browser faster than ASC2
24	1	array.join	29.60	4.93	4.93		0.20	29.60
25	2	array.sort	0.32	2.90	1.45		3.10	2.90
26	3	string concat	0.02	0.02	0.01		175.68	0.02
27	4	long concat (too inconsistent)	#DIV/0!	#DIV/0!	#DIV/0!		#DIV/0!	#DIV/0!
28	5	harmonic (flops)	0.77	0.88	0.62		1.61	0.88
29	6	substring	2.44	2.75	1.69		0.59	2.75
30	7	string indexOf	1.92	1.05	1.05		0.96	1.92
31	8	Math.round	2.86	3.10	0.06		17.47	3.10
32	9	empty loop	1.35	1.35	1.75		0.74	1.75
33	10	func + string + array	2.70	2.70	1.80		0.56	2.70
34	11	md5 cupcake	0.95	0.87	0.33		3.08	0.95
35	12	func call with return	1.00	1.00	0.13		7.65	1.00
36
37
38
39
40		Note: browser results were pretty inconsistent, jumping a lot more wildly than FP, so take that into account (sometimes it would go from 60ms to 16ms to the weird 2s)
41		Note: sometimes the browser (e.g. Firefox on #8) would hang for 2 seconds on the same test that would take 60ms otherwise, I have no idea what would cause that; in Firefox it even repeated several times in a row until I refreshed
42
43	Conclusion
44	From the small very artificial test set above:
45	Browsers seem better at		automatic inlining, string functions and optimizing loops (or general overhead when looping)
46	AVM seems better at		possibly string concatenation, lots of raw number crunching (Vector + uint) and flops (floating point operations)
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100