Another C++ entry
dave at thesmithfam.org
Wed Mar 22 21:55:50 MST 2006
Sasha Pachev wrote:
> Anybody wants to shake his C/C++ muscle?
My muscles are pretty puny, but here's my attempt in C. It runs in 1.9
seconds on my 2.4GHz RedHat box on the kjv100 file. I believe this is
one of the fastest times. The algorithm uses a constant-time hash table
of my own design, which lives in a very sparse array of 1 million
(mostly empty) entries. It runs with just 99 collisions in the hash's
key space. Interestingly, it turns out that I/O is not the big part of
the time (0.2 seconds to load both files). The fun part is that I didn't
do any string copies from the original data. That was kind of groovy.
The con is that it uses a TON of memory. It loads both files completely
into memory before working on the data. I could have worked around this,
to allow arbitrarily large files, but I've spent too much time now. :)
Bryan, could you run this code on your benchmark box? I'm curious how it
compares with the others.
P.S. To whomever it was that was whining about not having incentive:
Don't be a wimp. The code is the incentive.
P.P.S. I am NOT looking for employment. I already have the best job in
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
More information about the PLUG