From: Kirk Haines <khaines@...>
Date: 2005-10-14T02:03:44+09:00
Subject: Re: ANN:  Crypt::ISAAC 0.9.1 released

On Thursday 13 October 2005 10:50 am, Daniel Berger wrote:
> I got some improvement when I replaced "read" with "sysread" in ISAAC.rb.
> Here's a little benchmark script I wrote:

I'll make that change.

> bm do |x|
>     x.report("rand"){
>        MAX.times{ Crypt::ISAAC.new(true) }
>     }
>
>     x.report("/dev"){
>        MAX.times{ Crypt::ISAAC.new(false) }
>     }
> end

The default is true, which means to use the nonblocking entropy source 
(/dev/urandom), and while false tells it to use /dev/random.  It only uses 
rand() if it can't find the /dev/urandom (or /dev/random if told to use 
that).  So your timings that you have labeled "rand" are using /dev/urandom, 
and the one labeld "/dev" is using /dev/random.

> # Using read
>
>  >ruby bench_isaac.rb
>
>        user     system      total        real
> rand  6.290000   0.410000   6.700000 (  6.928163)
> /dev  6.490000   9.640000  16.130000 ( 23.197032)
>
> # Using sysread
>
>  >ruby bench_isaac.rb
>
>        user     system      total        real
> rand  6.340000   0.550000   6.890000 (  7.148228)
> /dev  6.560000   3.050000   9.610000 (  9.976435)

I'm surprised it was that fast, actually.  On my boxes, I had FAR less entropy 
in /dev/random available to me.  I actually thought that something was wrong 
with my code the first time I tested it, it was so slow.  :)

Thanks for the feedback,

Kirk Haines