The eIQ machine learning (ML) software development environment for
i.MX RT crossover MCUs supports the Glow
machine learning compiler,
which enables ahead-of-time compilation. The compiler converts the neural networks into object files, then
the user converts this into a binary image for increased performance and smaller
memory footprint as compared to a traditional runtime inference engine.
Glow is used as a software back-end for the PyTorch machine learning framework, including support for the ONNX model
format.
Glow, or graph lowering, compiler derives its name because it lowers a neural network into a two-phase strongly
typed intermediate representation. In the first phase, the optimizer performs
domain-specific optimizations. The second phase allows the compiler to perform optimizations that take advantage of
specialized back-end hardware features. It’s in this second phase that NXP has
added specialized support for Arm® Cortex®-M cores and Cadence®
Tensilica® HiFi 4 DSP support, accelerating performance by utilizing Arm CMSIS-NN and HiFi
NN libraries, respectively.
