Content deleted Content added
Line 45:
I updated this section to match the current link, Microsoft has changed their position on benchmarking should someone want to try to find references, seems they are mostly comparable from what I can tell.
I removed this section. Although I believe this is HIGHLY relevant to a comparison between the two and would love to see someone put it back in with a more thought out section. I thought it suffered from the following problems.
1 - The style and content of the opening paragraph was not up to par. This statement came off as pretty vacuous: ''"Comparing language speed is tricky. It depends on the virtual machine and the environment used. It can also depend on whether the benchmark code was written in an idiomatic or an optimised style."''
2 - The benchmark tests given are not useful. Java and C# are such similar languages that it should have been possible to directly compare them by essentially writing the same program line for line. However the algorithms/code used in these benchmarks differs substantially (for example, encoding things as bytes versus unicode characters, writing multi-threaded versus single threaded applications, optimization switches, etc.). The largest difference between Mono/Java (Binary trees test) is entirely an artifact of mono being run with an outdated version of the garbage collector, and the more common microsoft implementation shows essentially no difference at all for that test.
I would love to see a more rigourous section developed here. This article: http://www.itu.dk/~sestoft/papers/numericperformance.pdf seems like it could be a good start, though it only focuses on numeric computations, and it seems comparing garbage collectors, native interopt, memory usage, etc. would all be important.
Unless the article can say something meaningful, well supported, and useful, I think this section should remain out.
== Page Is Irrelevent ==
| |||