Threading with openResults() command

I am planning to have a threading of getting data from many psf Result
folder.

I am using Tcl/Tk and SKILL.
Now from Tcl/Tk Threading, I have distributed the Simulation Result
Extraction process into four local machine.

Example:
Local Machine:4
psf folders: /SimResult/test002 /SimResult/test003 /SimResult/test004

I just want to know if it is possible?

Because I'm thinking that since every time I used the openResults()
command, the results will be changed for the whole process. So the
value that has been extracted by getData() will be from only one psf
result folder.

Is there a way to make a threading in extracting data from psf result
folder?

I hope anyone could help me.
Thanks and God bless!

 

ecnedad wrote, on 01/22/10 08:07:

Virtuoso is single threaded, so the only way to do this is to run multiple
virtuoso/ocean executables, and do the openResults() in each process - that way
they will be independent.

Regards,

Andrew.

 

Thank you for the information.
We have make multiple ocean executables by the following:

myOceanScript.ocn -> script that extract the result using getData
("net013" ?resultsDir "result/psf" ?result 'dcOp).

We have executed the ocean executables using ipcBeginProcess:

cmd= "ocean -nograph <myOceanScript.ocn> result.log"
pid=ipcBeginProcess(cmd)

And we have noticed that the single openResults() and four getData()
process is faster that having parallel execution of four
myOceanScript.ocn.

Is there any faster way to extract result than using openResults() in
single process?

 

ecnedad wrote, on 01/27/10 03:44:

Without knowing what you're doing in the script in more detail, and the timings
of how long each part takes, it's very hard to know what to optimize.

For example, the cost of launching ocean may be more than your actual script
takes to run - I've no idea what the relative times are in your case.

If you're just accessing a few dcOp results, I can't imagine that there's any
benefit in doing it in parallel.

If you give more info, I (or others) might be able to give more advice.

Regards,

Andrew.