| Re: Compiling file with many functions: huge performance difference between IDL and IDLDE [message #38646 is a reply to message #38531] |
Thu, 18 March 2004 12:40   |
Sidney Cadot
Messages: 7
Registered: October 2002
|
Junior Member |
|
|
Oliver Thilmann wrote:
> Hi,
>
> your example is just generic for the kind of problem you want
> to solve, I assume. Otherwise why not use a hash? A very
> simple implementation (unsorted arrays) on a Pentium IV,
> 2.6 GHz, IDL 6.0 yields
>
> Setting 12000 random values (key: string, value: integer):
> mean 0.15 ms per entry (total ~2 seconds)
> Random access of 12000 values from this set:
> mean 0.3 ms per access, (total ~3.5 s)
>
> Is the access via call_function much faster?
I'm afraid to sound terribly stupid here, but is there support for
hashing in IDL? I haven't been able to find it.
My laborious trick is nothing more than circumventing the lack of
hashing as a standard feature in IDL (by piggybacking on the internal
hashing IDL uses for function names). If you know a better way, I would
be very much interested!
Best regards, Sidney
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
