Hello,
I am working on a project that utilizes NXP's i.MX 8M Plus Quad chip. I am exploring the best ways to leverage the GPU of this chip for general-purpose parallel computations. I have a few questions regarding this:
I am open to any suggestions to achieve the best performance for my project. Thanks in advance to everyone who can assist!
Best regards.
Please refer the Machine Learning Guide: https://www.nxp.com/docs/en/user-guide/IMX-MACHINE-LEARNING-UG.pdf
1.We recommend you use tflite model.
2.No, we add TFlite delegate support about NPU, but we don't provide any API like CUDA.