[Truffle] Eliminating Calls to Side-Effect Free Methods?

[Stefan Marr]

Hi:

On 12 Mar 2014, at 16:57, Stefan Marr <java at stefan-marr.de> wrote:

> At the moment, I am just implementing the evaluateVoid(.).
> Thinking about it, it seems to be a good way to propagate at least within a method, that all side-effect free executions can be ignored.
> 
> Let’s see what that gives.

Looks like Smalltalk just has a too small granularity of methods, and not really the constructs to get big benefits out of executeVoid().
Got only one benchmark showing a little speedup, the rest remains unchanged.
I guess, it could still help the partial evaluator but that would mean, I would need callVoid(.) and related to propagate the executeVoid calls through the method calls.


Best regards
Stefan

-- 
Stefan Marr
INRIA Lille - Nord Europe
http://stefan-marr.de/research/