[Pyrex] Re: any ideas to speed this up?

[Toby]

Brian Blais wrote:
> t1=time.time()
> ...
> print time.time()-t1

You might want to use time.clock() for benchmarking algorithms.  

time.time() measures the current (wall clock) time, so the difference
you print includes any time spent by the CPU on other processes, or
waiting for I/O, or responding to other computers on the network.

Instead time.clock() only measures the time spent by the CPU executing
your user code, discarding anything else.  There's a catch: on most
platforms time.clock() has a worse resolution (on my Linux pc it's .01s)
So if you need to time algorithms that take much less than a second, you
will have to repeat them a number of times, like this:

	t = time.clock()
	for i in range(1000):
	    # stuff <<1s to measure goes here
	print (time.clock() - t) / 1000


Toby

-- 
Signed/encrypted mail welcome.  GPG/PGP Key-Id: 0x15C5C2EA