| Re: strange behaviour of bytscl by large arrays [message #80045 is a reply to message #79953] |
Mon, 23 April 2012 12:00   |
Klemen
Messages: 80
Registered: July 2009
|
Member |
|
|
Lajos, thank you, this is the same thing as I observed, but it's nice to know that somebody else hase similar experience. You mentioned multhreading - is it possible, that the example I showed (it looks like having 4 areas - waves first and then 3 more constant areas) are due to my processor having two cores (four threads)?
David, sorry that I wasn't so clear in my first post. Yes my problem is, that I have 5 × 7523 × 11727 pixel large dataset. All the values are long integers, relativelly small (not more than 10 000). So the values are not the problem. But coverting my array to byte made me really question what the hell is going on. Conversion worked well on the subset but on the whole image not. The same is converting large float arrays to byte.
So can it be that some IDL functions don't work proper because they include FINDGEN in their code? Or some similar bug? I have today observed strange perfomance on large arrays using BYTE, BYTSCL, HIST_EQUAL, and also the belowed HISTOGRAM. Working on the doubble precision data seems not to be affected.
Cheers, Klemen
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
