Product Info

I replaced a different company's GPU with SiMa.ai! [Preparation]

SiMa.ai (hereinafter referred to as SiMa) is a device with strengths in edge AI, rivaling the performance of Alpha Inc., a company that provides GPUs (graphics processing units). There are also voices expressing interest in actually using SiMa's MLSoC (Machine Learning System-on-Chip). In this article, we will introduce how to port a model that was running on Alpha Inc.'s GPU to SiMa and actually run it.

Overall Table of Contents

The transplant procedure is as follows:

The goal of this chapter, "Preparation"

• Understand the environment required to run SiMa

The results of this demo video are as follows:

Table of contents for this chapter "Preparation"

1. What you need:
2. Overall picture of the transplant process

This section describes the preparations required before starting development and evaluation.

What you need:

Overall picture of the transplant process

Introduction

Development on SiMa mainly involves creating AI/ML applications on the host machine.
The host machine is where you do all the work of developing, deploying, and debugging your applications using APIs, frameworks, libraries, CLI tools, etc.
After developing an application, you deploy it from the host machine to the SiMa device and actually run the application on SiMa. The basic development flow is as follows:

1. Setting up the SiMa development board environment
2. Host machine environment setup
3. Building a heat map analysis system in SiMa
4. Build and deploy

Setting up the SiMa development board environment

Update your development kit firmware to the latest version.

Host machine environment setup

To set up the host machine environment, we will install an application provided by SiMa called Palette.
The application provides a single source of truth for developing, deploying, and debugging AI/ML applications with access to APIs, frameworks, libraries, and Command Line Interface (CLI) tools.
Also, since Palette runs in a Docker container, you will also need to set up a Docker environment.

In order to install Palette, you must meet all of the following installation requirements, so be sure to check them before starting development.

This chapter explains how to operate on Linux.

Requirement Type	detail
Host Machine	Architecture: x86_64 machine Number of CPU cores: Minimum 4 cores Memory: Minimum 16GB RAM Storage: At least 30GB of local storage [Recommended] SSD storage for faster installation
OS	Ubuntu: 20.04 LTS, 22.04 LTS Windows: 10 Pro, 11 Pro
Docker Engine	Ubuntu Docker Engine: v20.10.21 or higher Windows Docker Engine: v23.0
Python	Version: Python 3.8.10 or later
Open Port	Port 80 (HTTP)
Firewall settings	Port 80 (HTTP): Incoming and outgoing traffic Docker external network communication Access to remote devices via SSH using port 22

Building a heat map analysis system in SiMa

A system that uses Yolov7 to track people's movements and perform heat map analysis of their stay status will be run on SiMa via the host machine.
We use Gstreamer, a framework that makes it easy to process multimedia data.

Build and deploy

Connect to device and build^*1, Deploy^*2This is done using a command set called the MPK tool provided by SiMa.

---

In the next chapter, we will introduce how to set up the development environment on the development kit and host machine.