|
Re: Pyrex v Psyco: msg#00026python.pyrex
Philip Smith wrote: > I can't seem to replicate the performance > enhancement over Psyco which the Pyrex test suite demonstrates If you're talking about the primes example, keep in mind that it's a particularly good case, because it uses C operations exclusively in its inner loops. It's really a C program with a bit of Python around the outer edges. > Am I (as a newcomer) doing soething obviously wrong or is there > something about this code that means one wouldn't expect particularly > fast execution??? You're making a lot of Python API calls -- you can't expect spectacular performance from that. Also, you're calling the Python ord() function in a few places. That costs a dict lookup, a Python function call, and possibly some argument conversions. If you know you have a string (not unicode), you can do that a lot more efficiently by coercing to a char * and looking at the first character. -- Greg |
|
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| Previous by Date: | Pyrex v Psyco: 00026, Philip Smith |
|---|---|
| Next by Date: | Re: Pyrex v Psyco: 00026, John Machin |
| Previous by Thread: | Pyrex v Psycoi: 00026, Philip Smith |
| Next by Thread: | Re: Pyrex v Psyco: 00026, John Machin |
| Indexes: | [Date] [Thread] [Top] [All Lists] |
| News | FAQ | advertise |