Difference between revisions of "Download and setup LIBISIS"

From LIBISIS
Jump to navigation Jump to search
Line 1: Line 1:
 
* Download the latest libisis build [[Media:libisis.zip|libisis.zip]]
 
* Download the latest libisis build [[Media:libisis.zip|libisis.zip]]
 
* Unpack this file into <tt>'''C:\mprogs'''</tt> or a directory of your choice eg <tt>'''X:\analysis'''</tt>
 
* Unpack this file into <tt>'''C:\mprogs'''</tt> or a directory of your choice eg <tt>'''X:\analysis'''</tt>
 +
== Setup Paths ==
 
* Copy the libisis_start.m file into <tt>C:\mprogs</tt> or into the directory where your matlab starts
 
* Copy the libisis_start.m file into <tt>C:\mprogs</tt> or into the directory where your matlab starts
  
Line 11: Line 12:
 
>>
 
>>
 
>> libisis_start('X:\analysis\libisis_2008_01_18_1356')
 
>> libisis_start('X:\analysis\libisis_2008_01_18_1356')
 +
</pre>
 +
 +
== Setup Memory Usage ==
 +
 +
Sometimes when we are treating large datasets, such as on MAPS, we cannot load all the data into the memory of the computer at once, we need to 'chunk' the data into smaller pieces. We therefore need to set the default number of workspaces in each 'chunk' to treat at any one time during the population of an [[IXTrunfile]] and the total amount of memory to allow the computer to use.
 +
 +
If the memory required for the number of workspaces in each chunk is greater than that defined then the number of workspaces is reduced unitl it fits in the memory or a failure status is raised.
 +
 +
These two values are defined in <tt>./matlab/homer/default_homer_nchunk.m</tt> which needs to be edited to suit the memory needs of your computer. Tests have shown that LIBISIS will populate data fastest when it is chunked in groups of 100 workspaces, so this is the default value in the file.
 +
 +
<pre>
 +
function nchunk=default_homer_nchunk
 +
% nchunk = default_homer_nchunk
 +
% nchunk is a two membered array
 +
% the first element is the default number of workspaces to be treated at
 +
% one time, and the second element is the maximum amount of memory (in MB)
 +
% that you want the computer to use
 +
nchunk=int32([100,500]);
 
</pre>
 
</pre>

Revision as of 11:22, 11 February 2008

  • Download the latest libisis build libisis.zip
  • Unpack this file into C:\mprogs or a directory of your choice eg X:\analysis

Setup Paths

  • Copy the libisis_start.m file into C:\mprogs or into the directory where your matlab starts
  • there are two ways you can make sure the paths are set in the matlab instance
  1. call the libisis_start function from the command line, with the name of the libisis directory which was unpacked see the example below
  2. or you can add the line below into your startup.m, the path to the libisis_start function must of course be in the matlab path, so you may have to add C:\mprogs or X:\analysis to the path in the startup.m already
>> libisis_start('C:\mprogs\libisis_2008_01_18_1356')
>>
>> libisis_start('X:\analysis\libisis_2008_01_18_1356')

Setup Memory Usage

Sometimes when we are treating large datasets, such as on MAPS, we cannot load all the data into the memory of the computer at once, we need to 'chunk' the data into smaller pieces. We therefore need to set the default number of workspaces in each 'chunk' to treat at any one time during the population of an IXTrunfile and the total amount of memory to allow the computer to use.

If the memory required for the number of workspaces in each chunk is greater than that defined then the number of workspaces is reduced unitl it fits in the memory or a failure status is raised.

These two values are defined in ./matlab/homer/default_homer_nchunk.m which needs to be edited to suit the memory needs of your computer. Tests have shown that LIBISIS will populate data fastest when it is chunked in groups of 100 workspaces, so this is the default value in the file.

function nchunk=default_homer_nchunk
% nchunk = default_homer_nchunk
% nchunk is a two membered array
% the first element is the default number of workspaces to be treated at 
% one time, and the second element is the maximum amount of memory (in MB)
% that you want the computer to use
nchunk=int32([100,500]);