Thanks for the help. Yes, I see now that gdbm is really superfluous. Not having used pickel before, I started from the example in "Python in a Nutshell" which uses anydbm. Using a regular file saves some time. By far, the biggest improvement has come from using marshall rather than cPickel. Especially for writing. gc.disable() shaved off another 20%. I can't use psyco because I'm on x86_64. And my goal is to demonstrate how Python can combine simplicity, readability, *and* speed when the standard library is leveraged properly. So, in this case, calling C or assembler code would defeat the purpose, as would partitioning into smaller arrays. And my processor is single core.
I started at about 7 seconds to just read in and sort, and at 20 seconds to read, sort, and write. I can now read in, sort, and write back out in almost exactly 5 seconds, thanks to marshal and your suggestions. I'll look into struct and ctypes. Although I know *of* them, I'm not very familiar *with* them. Thank you, again, for your help. -- http://mail.python.org/mailman/listinfo/python-list
