From: Lunatic Leo <lleoem@...>
Date: 2011-07-28T21:02:00+09:00
Subject: Re: Fiber performance in 1.9.2-p290 vs -p180

First, thanks for looking at the code.

Later, I played with that after my original post. But even execution
times (rather than
creation+execution) show significant speed up between p180 and p290 .

% rvm use 1.9.2-p180
% ./fiber_ring.rb 20_000 1
creation : 20000, 1,  15.990000   0.070000  16.060000 ( 16.041650)
execution: 20000, 1,   1.300000   0.030000   1.330000 (  1.355492)

% rvm use 1.9.2-p290
% ./fiber_ring.rb 20_000 1
creation : 20000, 1,   0.480000   0.060000   0.540000 (  0.534756)
execution: 20000, 1,   0.200000   0.030000   0.230000 (  0.252246)

It is still 1.35s versus .252s for execution. I am happy and all. I was
hoping someone could point out if it was GC or Fiber optimizations.
I am relatively new to Ruby and I find Fibers facinating. I once
played with continuation type code in a C++ eventloop; like
em-synchrony does.


Thanks,
-LLeo

-- 
Posted via http://www.ruby-forum.com/.