Get Vitis Libraries - 2022.2 English

Now we need to clone the Vitis Libraries into local path.

The Vitis Library is open source on Xilinx GitHub: https://github.com/Xilinx/Vitis_Libraries.

Assuming you are using $/home/project$ as working directory, then use following command to clone the Vitis Library repository into the working directory:

cd /home/project
git clone https://github.com/Xilinx/Vitis_Libraries.git

This will take a few seconds to minutes for downloading depending on network. After it is completed, you may browse into the sub folders to get familiar with the file structure.

Vitis_Libraries/
├── Jenkinsfile
├── LICENSE.txt
├── README.md
├── blas/
├── codec/
├── data_analytics/
├── data_compression/
├── database/
├── dsp/
├── genomics/
├── graph/
├── hpc/
├── quantitative_finance/
├── security/
├── solver/
├── sparse/
├── utils/
└── vision/

There are many different categories of libraries under the top level folder. In this lab, we are going to use the dsp library. So let’s enter the the sub-directory dsp, and you can find following directory structure.

dsp/
├── Jenkinsfile
├── L1/
│   ├── README.md
│   ├── examples/
│   ├── include/
│   ├── meta/
│   ├── src/
│   └── tests/
├── L2/
│   ├── README.md
│   ├── benchmarks/
│   ├── examples/
│   ├── include/
│   ├── meta/
│   └── tests/
├── LICENSE.txt
├── README.md
├── docs/
│   ├── Doxyfile
│   ├── Makefile
│   ├── Makefile.sphinx
│   ├── README.md
│   └── src/
├── ext/
│   ├── README.md
│   ├── make_utility/
│   └── xcl2/
└── library.json

This library is written in C++ and it contains three levels of abstractions.

• L1: Module level, it provides optimized hardware implementation of the core LZ based and data compression specific modules like lz4 compress and snappy compress.

• L2: Kernel level, a demo on lz4, snappy, zlib and zstd data compression algorithms are shown via kernel which internally uses the optimized hardware modules.

• L3: The software API level will wrap the details of offloading acceleration with prebuilt binary (overlay) and allow users to accelerate data compression tasks on Alveo cards without hardware development.