From: Roger Pack <rogerpack2005@...>
Date: 2008-09-25T06:08:11+09:00
Subject: Re: Hash optimization question

> Well indeed I would suggest to anyone having problems with performance
> of small hashes to have a look at Array#assoc and #rassoc. The only
> problem is I do not know up to which number of pairs this behaves
> well. I guess n<20 should be quite safe. Maybe I will find some time
> to benchmark this later.

So our goal is to see if Array#assoc is "about as fast as hash based 
lookup"


>>  a = {1 => 2, 2 => 3 }
>>  b = [[1, 2], [2, 3]]

>> Benchmark.measure { 1000000.times { a[2] }}
=> #<Benchmark::Tms:0x5d93dc @total=0.26, @cutime=0.0, @label="", 
@stime=0.0, @real=0.270273208618164, @utime=0.26, @cstime=0.0>


>> Benchmark.measure { 1000000.times { b.assoc 1 }}
=> #<Benchmark::Tms:0x5b2e1c @total=0.26, @cutime=0.0, @label="", 
@stime=0.0, @real=0.264580965042114, @utime=0.26, @cstime=0.0>

faster than a hash :)


>> Benchmark.measure { 1000000.times { b.assoc 2 }}
=> #<Benchmark::Tms:0x5e4408 @total=0.35, @cutime=0.0, @label="", 
@stime=0.01, @real=0.35266900062561, @utime=0.34, @cstime=0.0>

Quite a bit slower.

So with only 2 entries #assoc is slower.

Oh well, we do what we can.
-=R
-- 
Posted via http://www.ruby-forum.com/.