Hacker News new | comments | show | ask | jobs | submit login

The lack of benchmarks comparing Mono to the CLR (C# VM) is very frustrating. Apparently, the lack of benchmarks is due to Microsoft's license forbidding the publication of benchmarks without Microsoft's permission (common practice among commercial vendors?).

Fortunately, we have the Benchmarks Game and that shows Mono performing well [on synthetic benchmarks...] when compared to a variety of languages.

vs Java: http://benchmarksgame.alioth.debian.org/u64q/benchmark.php?t...

vs Go: http://benchmarksgame.alioth.debian.org/u64q/benchmark.php?t...

vs Python: http://benchmarksgame.alioth.debian.org/u64q/benchmark.php?t...

We have included Mono in our Framework Benchmarks project for some rounds now. However, we have recently set up a new hardware environment and we're not having any luck getting Mono installed on that environment [1]. It's simply not installing correctly for us. We're not Mono experts, so we're probably doing something wrong.

If anyone is interested in lending a hand, we'd appreciate it since we'd like for the Mono tests remain in Round 9. Thanks!

[1] https://github.com/TechEmpower/FrameworkBenchmarks/issues/78...

Thanks. It took a little while to find, but if anyone is interested in comparing Mono to the CLR for themselves (and likely do some work to adapt the system to Windows), here's where to start: http://benchmarksgame.alioth.debian.org/play.php#languagex

The measurements scripts provided on the benchmarks game website are written in Python 2.5+ and worked fine on MS Vista using:

- Python 2.5+

- Python Win32 Extensions (for win32 process monitoring)

- GNU Make for Windows

- GNU DiffUtils for Windows

Download bencher.zip from:


Can someone not just make some code that does the benchmarks, and publish that?

Then some other helpful person can pastebin the actual results.

I think if you run a recent Mono, with the sgen GC using the LLVM backend and turn off array bounds-checking performance is supposed to be pretty impressive. (I recall Miguel saying within 10-20% of C++).

Having said that, I've had no performance issues running 3.2 on Linux in general.

>>Microsoft's license forbidding<<

I seem to remember that changed; and for a couple of years the benchmarks game did provide measurements for C# F# (.Net and Mono) and Java on MS Vista.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact