more similar

When programs are idiomatic someone will say they are not comparable.

Are they similar-enough to be comparable for your purposes? Will someone then say the programs are not idiomatic.

If only more-idiomatic was not less-comparable!

Here are a few naive un-optimised single-thread #8 programs transliterated line-by-line literal style into different programming languages from the same original.

		Mean or 95% CI	insignificant io
×	source	cpu secs
1.0	Swift	34.68—34.74	fannkuch-redux
1.0	C gcc	5.54	n-body
1.0	Intel C	2.68	spectral-norm
1.0	C clang	5.70—5.75	n-body
1.0	C clang	36.38—36.47	fannkuch-redux
1.1	Intel C	6.06—6.10	n-body
1.1	C gcc	39.52—40.22	fannkuch-redux
1.2	Swift	6.56—6.95	n-body
1.2	Intel C	42.88—43.00	fannkuch-redux
1.3	C# naot	7.01	n-body
1.3	C# naot	44.63	fannkuch-redux
1.3	Dart aot	7.25	n-body
1.3	Java naot	7.29—7.33	n-body
1.3	Go	46.26—46.33	fannkuch-redux
1.3	Java	49.88—55.68	fannkuch-redux
1.4	Java	7.66	n-body
1.5	Java naot	51.94—51.98	fannkuch-redux
1.7	Node.js	58.31	fannkuch-redux
1.7	Node.js	9.64—9.99	n-body
1.7	Go	9.64	n-body
1.7	Dart aot	60.61	fannkuch-redux
2.0	C# naot	5.34	spectral-norm
2.0	Java naot	5.35	spectral-norm
2.0	Swift	5.35	spectral-norm
2.0	Go	5.36	spectral-norm
2.0	C gcc	5.41	spectral-norm
2.0	C clang	5.42	spectral-norm
2.1	Java	5.51	spectral-norm
2.4	Node.js	6.53	spectral-norm
3.1	Dart aot	8.38	spectral-norm
20	Java -Xint	709.01	fannkuch-redux
24	PHP	849.92—863.13	fannkuch-redux
29	Java -Xint	161.04—161.34	n-body
32	Java -Xint	84.24—84.42	spectral-norm
44	PHP	242.50—244.36	n-body
51	PHP	136.38—136.78	spectral-norm
58	Ruby	318.33	n-body
59	Ruby	2,058.52	fannkuch-redux
66	Ruby	175.87—176.32	spectral-norm
71	Python 3	2,455.59	fannkuch-redux
95	Python 3	528.16	n-body
144	Python 3	385.07	spectral-norm

Seemingly similar programs behave differently when the language implementations behave differently. Do they have the same default stdio buffering? Do the programs need to be written differently and run differently to make them behave the same.

To be less similar so they are more comparable!

		Mean or 95% CI	significant io
×	source	cpu secs
1.0	C# naot	6.30	fasta
1.0	Intel C	23.98	mandelbrot
1.0	Go	62.37—62.83	k-nucleotide
1.0	C gcc	25.18	mandelbrot
1.1	C clang	25.50—25.55	mandelbrot
1.1	C# naot	25.55—25.66	mandelbrot
1.1	Swift	25.60—25.64	mandelbrot
1.1	Go	26.77	mandelbrot
1.2	Java	27.85—27.91	mandelbrot
1.2	Java naot	28.53—28.69	mandelbrot
1.2	C# naot	75.78—77.28	k-nucleotide
1.2	Intel C	7.90—7.98	fasta
1.3	PHP	80.05—80.48	k-nucleotide
1.3	C clang	8.19—8.30	fasta
1.3	C gcc	8.31—8.37	fasta
1.3	Java	83.79—87.16	k-nucleotide
1.4	Java naot	87.80—89.42	k-nucleotide
1.7	Java naot	10.48—10.54	fasta
1.7	Node.js	109.35—113.96	k-nucleotide
1.7	Java	10.94—11.09	fasta
1.8	Go	11.54—11.80	fasta
2.7	Swift	168.68—173.38	k-nucleotide
2.9	Swift	18.18—18.35	fasta
2.9	Node.js	73.08—75.99	mandelbrot
3.0	Node.js	19.26—19.53	fasta
3.8	Python 3	235.36—236.89	k-nucleotide
4.5	Dart aot	28.61—29.04	fasta
7.5	Dart aot	181.69—186.69	mandelbrot
7.9	PHP	49.85—49.96	fasta
8.3	Ruby	517.57	k-nucleotide
9.1	Dart aot	565.05—569.06	k-nucleotide
18	PHP	422.73	mandelbrot
24	Python 3	153.78—158.43	fasta
26	Java -Xint	632.13	mandelbrot
30	Ruby	188.50—189.24	fasta
39	Ruby	943.02	mandelbrot
52	Java -Xint	327.85	fasta
66	Python 3	1,575.80	mandelbrot

Such comparisons raise a nest of issues all their own: for example, is it really the same program when written in the two compared languages? (Will Partain, 1992.)

 

Throw the ratios together and give them a shake!

cpu secs / fastest				Geometric Mean
1.1	Intel C
1.2	C clang
1.3	C gcc
1.3	C# naot
1.5	Go
1.5	Java
1.5	Java naot
1.6	Swift
2.2	Node.js
3.6	Dart aot
15	PHP
30	Java -Xint
36	Ruby
42	Python 3