Surprising (for me) benchmark results...

[E. Mark Ping]

In article <XFMail.20010501144036.shaleh at valinux.com>,
Sean 'Shaleh' Perry <shaleh at valinux.com> wrote:
>> 
>> start = time.time()
>> f = open("infile.txt", 'r')
>> out = open("outfile.txt", 'w')
>> l = f.readlines()
>> l.sort()
>> out.writelines(l)
>> f.close()
>> out.close()
>> end = time.time()
>> 
>> print (end-start), " seconds"
>> 
>> Python runtime:  600 ms
>> 
>
>realize that readlines() reads the entire file into memory, so the sort was
>100% in memory.
>
>On today's machines reading an entire 4.5 mb file is not that bad, but still
>something to remember.

Yes, that was one of the parameters of the test.  All of the
implementations used that method (reading the entire file first).  We
even tried a file 10x the size, and the results scaled quite well
(which made sense, since the IO dominated).
-- 
Mark Ping                     
emarkp at soda.CSUA.Berkeley.EDU