Accelerator#
To integrate the MX3 into your application, use the runtime library which we call Accelerator. It comes in Python and C++ variants.
As is typical with other software, the C++ API is slightly more performant (especially in multi-stream applications), but the Python API is more friendly for beginners or quick prototyping.
If you only want to benchmark a model’s performance, please see the Benchmark CLI Tool instead.
See also