Micro-benchmarks are statistics, and of course, you can angle your 
benchmark to tell whatever story you want. Yes, this stuff is subject to 
various tricks and short-circuits

OTOH, You can drown out anything with enough skepticism and if you are so 
emotionally entrenched, you can refuse to believe anything.

I believe my benchmarks were meaningful in a small way. If you refuse to 
believe them, I'm not going to labor an argument further.

the guys over here: http://shootout.alioth.debian.org/. put a *ton* of time 
into polishing these types of micro-benchmarks and still they aren't 
remotely perfect. I don't have the motivation to sink that much time + 
effort into that.

On Wednesday, October 3, 2012 11:23:29 AM UTC-5, Casper Bang wrote:
>
>
> I feel the same sentiment though in reverse. Why are so many people, even 
>> in a Java group, pointing out how much faster C# is with integer arrays, 
>> when benchmarks indicate the exact opposite?
>>
>
> I feel tempted to say something about the uselessness of micro-benchmarks 
> on a managed runtime, and yet somehow attracted to dive into the matter. 
> However before doing so, you should make multiple runs, cut away the best 
> 25% and the worst 25% and average the result. Then normalize your numbers 
> to make it comparable across platforms. Also, how about coming up with a 
> real problem less susceptible to compiler/JIT short-circuits, i.e. finding 
> primes via % and use array/list as cache? This would also allow Scala 
> enthusiasts start beating the concurrency drum.
>

-- 
You received this message because you are subscribed to the Google Groups "Java 
Posse" group.
To view this discussion on the web visit 
https://groups.google.com/d/msg/javaposse/-/7N9hcUFa0HEJ.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/javaposse?hl=en.

Reply via email to