I know it’s old, but the results are pretty good. I wonder how v1.0+ compares.
https://modelingguru.nasa.gov/docs/DOC-2676
The timing setup is not very clear, but I am under the impression that it includes compilation time.
If that is the case, I am not sure I would consider this comparison useful; running a single Fibonacci calculation per process may not be a common use case.
The report seems to be a little bit messy, e.g.
Problem 1: Fibonacci Number
Language Option n=25 n=35 n=45
R 0.034 0.034 0.034
The same for the recursive version with 0.008
Why is that?
The source code does not contain the R code! 
It would be so cool if there was a package that:
- checks which languages are locally available
- run a set of tests on those languages
- submits the results with all the other hardware stats to some repo
- and then some stats can get compiled on all the uploaded results.
Since this is something all the involved languages would be equally interested in, a joint effort might make most sense?
How would this be different from eg
?
No idea, didn’t know about that. Cool!
Really good points. Thank you.
I am guessing that R is probably using a pre-computed table.
Probably, but using this in a benchmark? Questionable! 