Background: I have been running a script locally on my computer that looks at a few hundred files but I am about to scale up the number of files to about 10,000 and my script scales as the square of the number of files, so my computation time on my four-core computer would take several weeks. So I want to run this on the Amazon cloud.
I did quite a bit of googling and reading up on the Amazon cloud, but I still had a few questions (technical and non-technical):
1. Is there an hourly rate for the Distributed Computing Toolbox (I have the Parallel Computing Toolbox but I would rather rent the Distributed toolbox since I imagine I would only be running this software for a total of a few hundred hours)?
2. The maximum number of workers for the Distributed Computing Toolbox appears to be 64 workers. Does this only apply to the number of CPU cores or does it also apply to GPU cores (for GPUs, 64 would seem to be a pretty low number). Right now, I haven't used any gpuArray elements so it would not be compatible with GPUs, but I think I could probably modify it pretty easily if needed.
3. For running scripts on the Amazon cluster, do I need to run the script from the command line? My .m file right now has a GUI and I would prefer to run the GUI, select all of my files interactively that I want to run along with several parameters, and then run. I'm okay with paying for the entire time the script is running (including the GUI time when the user is selecting files), but I was wondering if this was allowed and how the charges would be calculated (after the script finishes, do I continue to get charged until I exit my session)? I guess the more general question is, does Amazon charge as soon as you enter the cluster or does it charge only while the script is running (minus any time wall time when it's waiting for resources).