| Re: Maximum Likelihood processing time [message #38507 is a reply to message #38421] |
Wed, 10 March 2004 02:28  |
lbusett
Messages: 9
Registered: March 2004
|
Junior Member |
|
|
> Hello,
>
> I think what you are seeing is that when a routine is first called, IDL
> has to search for it and (possibly) compile it. Subsequent calls don't
> have to be compiled.
>
> You could compile the routines before running the routine (see .compile)
> or you could perform your comparisons only after the routine has been
> called at least once.
>
> Ben
Hi Ben,
thanks for your answer. Unfortunately, I dont' think that this is
the (only) problem. I'll try to explain what happens with an example:
- I run the classification for the first time with 5 bands, and I get
a processing time of 12 seconds. Now, all the routines have been
compiled, and, if your suggestion is correct, the processing time
should be constant for further classifications, but I noticed the
following problem:
- If, after the first 5 bands classification, I run a 50 bands
classification, the processing time is about 300 seconds.
- If, after the 5 bands classification, I run an intermediate 45
bands classification (which takes about 275 seconds), and then a 50
bands classification, the time required for the 50 bands
classification decreases to about 85 seconds !
- If now I quit IDL, restart it, and perform again the 50 bands
classification, I get a processing time of 15 seconds !
Now, I think that this behaviour could be related to a "memory
allocation" problem, but I don't know how to manage it.
Does anyone have a suggestion ?
Thanks for the help,
Lorenzo
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
