In the ReleaseNotes of NCSDK v2.04.00, there is a statement which says 'Multiple network graphs can be allocated to a single neural compute device'.
Does the statement mean we can call graph.allocate() for multiple graph files?? (I'm assuming graph files are the files that gets generated during mvnccompile). Is it valid to attempt queuing inference on such multiple graphs?
Hi @madhusudhan_s - yes, you can allocate multiple graph files to a single device with NCSDK 2.x. The graph.queue_inference() method will queue an inference using the graph object you called it with.
You can read more about using the new API on the Python API Overview documentation page: https://movidius.github.io/ncsdk/ncapi/ncapi2/py_api/readme.html.