<div dir="ltr"><div>Thanks for pointing that out. It was a mistake indeed.<br></div>Interestingly enough, changing it from (random 23) to (random 26) resulted in a slight increase in running time from 500 ms to 530 ms.<br>
</div><div class="gmail_extra"><br><br><div class="gmail_quote">On Fri, Jul 25, 2014 at 1:39 PM, Jake Eakle <span dir="ltr"><<a href="mailto:jseakle@gmail.com" target="_blank">jseakle@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">This is probably not responsible for much of the performance difference, but the racket code has (random 23) where it looks like it ought to be (random 26), resulting in a smaller space of possible random keys. This will cause the branch with the lookup to happen somewhat more often than in the python version.</div>
<div class="gmail_extra"><br><br><div class="gmail_quote">On Fri, Jul 25, 2014 at 1:34 AM, Laurent <span dir="ltr"><<a href="mailto:laurent.orseau@gmail.com" target="_blank">laurent.orseau@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div><div><div>Even more idiomatic:<br></div><div>(time (for ([w words])<br></div> (hash-update! d w add1 0)))<br>
<br></div>But it seems to be slightly slower that Robby's version.<span><font color="#888888"><br></font></span></div><span><font color="#888888"><div><br></div>Laurent<br>
</font></span></div><div><div><div class="gmail_extra"><br><br><div class="gmail_quote">On Thu, Jul 24, 2014 at 11:26 AM, Robby Findler <span dir="ltr"><<a href="mailto:robby@eecs.northwestern.edu" target="_blank">robby@eecs.northwestern.edu</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Not an answer to your direction question, but this is the more<br>
idiomatic way to write that and it seems to be a bit faster:<br>
<br>
(time (for ([w (in-list words)])<br>
(hash-set! d w (add1 (hash-ref d w 0)))))<br>
<div><div><br>
On Wed, Jul 23, 2014 at 10:54 AM, Pedro Ramos<br>
<<a href="mailto:pedropramos@tecnico.ulisboa.pt" target="_blank">pedropramos@tecnico.ulisboa.pt</a>> wrote:<br>
> Hi,<br>
><br>
> I've been developing an implementation of Python in Racket, where I'm<br>
> implementing Python's dictionaries over Racket custom hash tables.<br>
><br>
> While my occasional benchmarks typically show better performance on Racket<br>
> programs than their Python equivalents, Racket's hash tables generally seem<br>
> to be slower than Python's dicts.<br>
><br>
> I've set up this benchmark in Racket:<br>
><br>
><br>
> #lang racket<br>
><br>
> (define alphabet "abcdefghijklmnopqrstuvwxyz")<br>
> (define (random-word n)<br>
> (build-string n (lambda (x) (string-ref alphabet (random 23)))))<br>
><br>
> (define words (for/list ([k 1000000])<br>
> (random-word 3)))<br>
> (define d (make-hash))<br>
><br>
> (time (for ([w words])<br>
> (if (hash-has-key? d w)<br>
> (hash-set! d w (add1 (hash-ref d w)))<br>
> (hash-set! d w 1))))<br>
><br>
><br>
> And its equivalent in Python:<br>
><br>
><br>
> import random<br>
> import time<br>
><br>
> alphabet = "abcdefghijklmnopqrstuvwxyz"<br>
> def random_word(n):<br>
> return ''.join([random.choice(alphabet) for i in range(n)])<br>
><br>
> words = [random_word(3) for k in xrange(1000000)]<br>
> d = {}<br>
><br>
> a = time.time()<br>
> for w in words:<br>
> if w in d:<br>
> d[w] = d[w] + 1<br>
> else:<br>
> d[w] = 1<br>
> b = time.time()<br>
> print b-a, 'seconds'<br>
><br>
><br>
> The Racket example yields running times of around 500 ms (running on Racket<br>
> v6.0.1) while the Python example yields running times of around 330 ms<br>
> (running on Python 2.7.3).<br>
><br>
> I find this unusual because Python is somewhat more dynamic than Racket,<br>
> since<br>
> (a) Python's equality and hashing functions have to dispatched at runtime<br>
> for each key;<br>
> (b) referencing and setting values in a Python dict is done using a very<br>
> general operator, [], whose behaviour also has to be dispatched at runtime,<br>
> unlike the more specific hash-ref and hash-set! Racket functions.<br>
><br>
> Is there something I'm missing about Racket's hash tables which explains<br>
> this slower speed?<br>
><br>
> Thanks,<br>
> Pedro Ramos<br>
><br>
</div></div>> _________________________<br>
> Racket Developers list:<br>
> <a href="http://lists.racket-lang.org/dev" target="_blank">http://lists.racket-lang.org/dev</a><br>
><br>
_________________________<br>
Racket Developers list:<br>
<a href="http://lists.racket-lang.org/dev" target="_blank">http://lists.racket-lang.org/dev</a><br>
</blockquote></div><br></div>
</div></div><br>_________________________<br>
Racket Developers list:<br>
<a href="http://lists.racket-lang.org/dev" target="_blank">http://lists.racket-lang.org/dev</a><br>
<br></blockquote></div><span class="HOEnZb"><font color="#888888"><br><br clear="all"><div><br></div>-- <br>A warb degombs the brangy. Your gitch zanks and leils the warb.
</font></span></div>
</blockquote></div><br></div>