Slower than Python

[SomeDude]

On Sunday, 3 March 2013 at 11:57:36 UTC, bearophile wrote:
> Walter Bright:
>
>> and I have trouble thinking of any significant improvements to 
>> code gen that could be made with runtime data.
>
> I've seen that the JavaHotSpot is able to unroll loops with a 
> length known only at run-time. It unrolls them only when the 
> run-time statistics say it's advantageous.
>
> (In theory the same is possible with profile-driven 
> optimization).
>
> Bye,
> bearophile

I believe the Hotspot runtime compiler is about the only one to 
perform runtime optimizations like that. I don't know if the CIL 
does this kind of things.