[i.MX 8X]Some questions about hardware acceleration of AI models

[ShenYi]

Hello，

I want to deploy the onnx model on i.MX 8X to perform face detection. But I found that i.MX 8X is not mentioned in the eIQ ML Software Development Environment.

So,

(1) Can the GPU of i.MX 8X accelerate the inference speed of AI models? Is there any reference documents?

(2) Can GPU or A35 of i.MX 8X support ONNX RUNTIME?

(3) Does i.MX 8X support performance optimization for OpenCV?

Expect for your reply.

[pengyong_zhang]

Hi @ShenYi 

Sorry, We do not have tested it, But from openvino, OpenVINO™ supports inference on CPU (x86, ARM), GPU (OpenCL capable, integrated and discrete) and AI accelerators (Intel NPU). So i think maybe you can try it on your site. This is not an area we support.

B.R

[pengyong_zhang]

Hi @ShenYi 

(1) Can the GPU of i.MX 8X accelerate the inference speed of AI models? Is there any reference documents?

>>> No, i.MX8X GPU do not support this.

(2) Can GPU or A35 of i.MX 8X support ONNX RUNTIME?

>>>GOU does not support, CPU can run it.

(3) Does i.MX 8X support performance optimization for OpenCV?

>>>Yes, i.MX 8X support OpenCV

B.R

[ShenYi]

Hello,
Thanks for your reply.
Another question is whether A35 support OpenVINO?