Installing TVM with Arm Ethos-U support
In the previous recipe, we briefly talked about the Ethos-U55 program, a command stream used to instruct the operations to execute on the microNPU. However, how is the command stream generated? In this chapter, we will be using TVM, a Deep Learning (DL) compiler technology that aims to generate C code from an ML model for a specific target device.
In this recipe, we will learn what TVM is by preparing the development environment that we will use later on in the chapter.
Getting ready
The goal of this recipe is to install the TVM compiler from the source. The installation needs the following prerequisites:
- CMake 3.5.0 or later
- C++ compiler with C++14 support (for example, g++ 5 or later)
- LLVM 4.0 or later
- Python 3.7 or Python 3.8
Before getting started, we recommend that you have the Python virtual environment (virtualenv) tool installed to create an isolated Python environment. You can refer to Chapter 7, Running...
