...
The MetaShape application does appear to be able to utilize multiple cores as various processing stages are detailed as being parallelized.
GPU Enabled
MetaShape is GPU enabled. Currently this has not been tested via any form of batch job. Also be aware that interactive desktops currently do not have a GPU attached to them.
For further details please consult the manual, and be aware of the --gpu-mask / --cpu-enable flags detailed when running metashape -h.
Multiple Nodes / Distributed Processing
MetaShape can be configured to run across a cluster where “multiple Metashape instances running on different processing nodes can work on the same task in parallel, effectively reducing the total processing time needed.“. Currently this has not been tested - but ARCC is looking to work with a group to test how this functionality can be setup and used.
GPU Enabled
MetaShape is GPU enabled. Currently this has not been tested via any form of batch job. Also be aware that interactive desktops currently do not have a GPU attached to them.
For further details please consult the manual, and be aware of the --gpu-mask / --cpu-enable flags detailed when running metashape -h.
Python/Java Development
The MetaShape stand alone Python module/wheel and Java api jar can be downloaded from https://www.agisoft.com/downloads/installer/ - further details can be found within the pro manual.
To install the Python wheel, you will have to install using explicitly use the explicitly shipped version of Python with the version of MetaShape you are using. For example:
...
Also note that you can perform Python development within MetaShape’s own console and/or run python scripts using the metashape -r <script> flag.