AMD Ryzen AI NPU: Image Zero Padding

OverView

This session demonstrates a zero padding of an image using the Multi dimension DMA capabilities of AIE shim DMA of AMD Ryzen AI Phoenix using AIE DIALECTS and AIE API.

Requirements

AMD Ryzen AI Phoenix.

Linux® based development environment

Python® (for test automation and result validation)

IRON API and MLIR-based AI Engine Toolchain

OpenCV (Open Source Computer Vision Library)

Project Brief

The SOC is designed to accelerate AIE-ML algorithms, delivering good to exceptional performance. The NPU has complex

• 16 AI Cores for computation
• 4 Memory Tiles for fast memory access
• 4 SHIM DMA to Move data in and out of L3 Memory

Note: This project is customized for Phoenix.

Features covered in this session

How to perform a zero padding of an image using the shim DMAs of “AMD Ryzen AI NPU”.

Architecture

Multidimensional DMA

Add reference to

https://www.hackster.io/ajoejose/amd-ryzen-ai-npu-multi-dimensional-shim-dma-9b8afd

Zero padding

In image processing and deep learning, zero padding is the process of adding a border of pixels with a value of 0 (which appears black) around the edges of an image.

Why Use Zero Padding?

• Maintain Image Size: Without padding, sliding a filter (kernel) over an image shrinks the output dimensions. Zero padding allows the output to match the original input size.
• Preserve Edge Data: It allows the filter to process the pixels at the very edges of the image as thoroughly as those in the center, preventing information loss.
• Computational Efficiency: In Signal Processing (like FFT), padding to a power of two can speed up calculations.

Note: Image dimensions must be in multiples of 8 pixels.

Padding Configuration

K = width of the image

M = height of the image

Kl = Padding Left

Kr = Padding Right

Mt = Padding Top

Mb = Padding bottom

offset=(Mt*(K + Kr + Kl) + Kl)

sizes=[1, 1, M, K]

strides=[ 1, 1, K + Kr + Kl, 1]

Input

Output

How to Build and Run

To compile the placed design:

env use_placed=1 make
make zero_padding.exe

To run the design:

./zero_padding.exe -x build/final_256.xclbin -i build/insts.bin -k MLIR_AIE --image input_image.jpg

How to Install and Set Up the Environment?

https://www.hackster.io/541340/amd-ryzen-ai-npu-tool-chain-installation-and-execution-b252fa

Conclusion

This session demonstrates how to use the “IRON API and MLIR-based AI Engine Toolchain” and “AIE API” to perform a zero padding of an Image.