Calling it with fsl_sub is not what you want to do.
You might try copying randomise_parallel to your own directory and editing
the two calls to fsl_sub to specify the full path of your modified version
of fsl_sub.
Cheers,
-Adam Thomas
On 8/22/12 8:53 PM, "bettyann" <[log in to unmask]> wrote:
>I am going to move ahead and work with calling 'fsl_sub
>randomise_parallel' directly.
>
>If anyone feels this is not the correct way to run randomise on a
>multi-node cluster, I would appreciate some correction.
>
>Thanks,
>- BettyAnn
>
>
>> Hello all,
>>
>> Could someone help me understand the mechanics of running randomise in
>> a parallel fashion?
>>
>> I am running on a PBS cluster. I have copied and modified fsl_sub in
>> my own directory. How do I run randomise with my PBS-modified fsl_sub
>> script?
>>
>> Since randomise is 'auto-aware' of the cluster, does it just 'find'
>> fsl_sub? If so, how? How will it find my PBS-modified version?
>>
>> Or do I need to explicitly run randomise_parallel thru fsl_sub like
>> so:
>>
>> ./fsl_sub randomise_parallel -i <input> ...
>>
>> After searching around, I am still confused. I don't understand who
>> calls what in what order.
>>
>> Thanks,
>> - BettyAnn
|