benchmarks
benchmarks copied to clipboard
kostya
Suggestions
Please include C Clang for the Base64 benchmark. My results on my machine: GCC: encode: 1333333600, 1.08 decode: 1000000000, 2.07
Clang: encode: 1333333600, 1.23 decode: 1000000000, 1.44
Also, please modify the D implementation of the Matmul benchmark. dotProduct is optimized, every other language usees the naive implementation (that's why D is so fast in this benchmark).
i little update c test, and right now gcc is faster (my clang clang version 3.5.1-svn225255-1~exp1)
I'm a little curious if you have any plans to do some testing where the startup/shutdown time and vm warming are not included in the test time. Clearly this would help languages like Java/Scala/Clojure.
Add table with an average of each language of the overall rank results would be interesting
@joaoeudes7 Unfortunately, the current tests can't be used for the overall ranking as the most of them is about libraries (standard or 3rd parties), not the languages itself. For example, matrix multiplication is extremely slow in Python in general, but magnitude faster when utilizing numpy. However, nobody is going to do matrix math in Python without numpy or the similar libraries, so ranking it as a slow for matrix math would be slightly incorrect.
The only test measuring the actual languages performance is bf2 (havlak too, but the number of supported languages is much smaller, and it's a candidate for the complete removal from the suite). So you may consider this one as the "overall rank results". If you may want more comprehensive results, I'd like you to refer to Debian benchmarks game.
Here we rather have a playground as it doesn't require much efforts to write tests, and some exotic languages could be verified. Debian-like tests would require much more efforts and contributing time which we don't have, sorry.
No problem, what you said really makes you feel in that context where the tests are. Thanks for clarifying