Attempt to unroll stuff. Part of this is making space.newbool be branch ...

[alex]

...free

[alex]

Notably missing is any sort of cap on how many times it will unroll (can add later)

[alex]

Also:

i = 0
while i < 10000000
  i += [0, 1].reduce(:+)
end

doesn't actually seem to be any faster than before this. So in conclusion, wat :/ (Good news: looking at the trace, it looks like maybe at least something got unrolled)

[timfel]

So, I think this looks good, although it's strange that it doesn't actually speed things up. For the pixels benchmark, do you think we should refactor to use the rlib ordered dict instead of rolling our own?

[alex]

That would probably be a good idea. I also think I should split this patch
into the 2 parts; the bool refactor and the unrolling.

On Wed, Aug 6, 2014 at 2:08 AM, Tim Felgentreff notifications@github.com
wrote:

So, I think this looks good, although it's strange that it doesn't
actually speed things up. For the pixels benchmark, do you think we should
refactor to use the rlib ordered dict instead of rolling our own?

—
Reply to this email directly or view it on GitHub
#836 (comment).

"I disapprove of what you say, but I will defend to the death your right to
say it." -- Evelyn Beatrice Hall (summarizing Voltaire)
"The people's good is the highest law." -- Cicero
GPG Key fingerprint: 125F 5C67 DFE9 4084