| 24 Aug 2012 03:06 AM |
|
|
Could it be a problem to handle with 3 dimensional arrays in IDL? It seems to me that IDL have big problems with such big arrays and so the calculation time for such arrays is extremely high. Are there possibilities to handle with such arrays for instance with the help of GPULib?
best regards,
peter
|
|
|
|
Deleted User
New Member
Posts:  
| 27 Aug 2012 12:06 PM |
|
|
IDL is really good at working with arrays of many dimensions. If you experience longer processing times it could be a limitation of your operating system or the code you are using isn't as efficient as it could be. There are quite a few routines that take advantage of 'multi-threading' which allows IDL to split the processing among the different cores of your operating system. The IDL Help can provide the list of those functions.
1) What are your array dimensions and what is the data type?( float, byte, integer, etc...)
2) What calculations are you using and how long do they take?
I'm not familiar with the GPILib and how it works with IDL. Maybe their website has more information.
|
|
|
|
Deleted User
New Member
Posts:
| 30 Aug 2012 01:24 AM |
|
|
First of all thank you for your help.
array dimensions: 1000x1000x1000. Data type: float
I am doing mathematical operations on the arrays like: exp, cos, sin and simple operations like matrix multiplication and addition but I also use transpose, rebin.
The dimension of the arrays are big, so I think that I could also get a problem with the RAM (48GB)?
At the moment I am working on the efficiency of the code.
|
|
|
|
Deleted User
New Member
Posts:
| 30 Aug 2012 08:58 AM |
|
|
[QUOTE]heftber wrote
The dimension of the arrays are big, so I think that I could also get a problem with the RAM (48GB)?[/QUOTE]
Ya think!
|
|
|
|