I recently made batch scripts for running multicore CP jobs on 2, 4, or 8 core workstations. The batch script appropriate for your number of CPUs should work without requiring much user modification after running the pipeline containing a “CreateBatchFiles” module at the end. The batch script does a little integer math to figure how to equitably split up the processing load by counting the number of images named “Blue.tiff” since split RGB image files are needed for the downstream CPA classifier module. An image batch job that contains an odd number of images will process the extra image with core1. The batch file also assumes that the measurement table will by copied into the typical CP image sub-directory labeled “output”. I think that a batch file could be wrote out from the “CreateBatchFiles” module just as is done with the properties file in the “ExportToDatabase” module if a “make multicore batch file script” checkbox option is implemented in the next version of CP2. I attached the 2, 4 and 8 core versions of the batch file to the post. Just unzip and copy the bat file into the image directory then click to run the image processing job on all cores available to you. By the way, the ExportToDatabase module does not robustly handle parallel sqlite writes from multiple processes, so, the MySQL/CSV or SQL output options have to be used with the current r10997 CP version. Hopefully this will be fixed in future CP updates since sqlite databases are the fastest measurement table storage format for reading data into the CPA classifier. Just a word of warning to anyone interested in running parallel CP2 image processing jobs.
8core batch script-drag to image directory and click.zip (574 Bytes)
2core batch script-drag to image directory and click.zip (472 Bytes)
4core batch script-drag to image directory and click.zip (511 Bytes)