User guide¶

To use the Python APIs provided here, please first follow the steps below to set up your environment

1. Set up Anaconda environment

Please follow the instructions described in Python environment setup guide to install anaconda2 and set up gemx environment. All test functions and examples should be run under gemx environment. Please deactivate gemx environment after testing.

2. Build library libgemxhost.so

In order to build libgemxhost.so, you need to either have Xilinx Runtime (XRT) installed.

$ pushd C++
$ make
$ popd

3. Set PYTHONPATH environment variable to point to gemx python bindings

$ export PYTHONPATH=./python

You can run the following command to verify if your environment is set up correctly.

$ python ./tests/test_gemm.py --xclbin ./xclbins/u200_201830_1/gemm_short/gemx.xclbin --cfg ./xclbins/u200_201830_1/gemm_short/config_info.dat --gemxlib ./C++/lib/libgemxhost.so

More details for testing and benchmarking GEMX Python APIs can be found in the User guide for testing and benchmarking GEMX Python APIs.

To find more about how to use GEMX Python APIs for offloading MLP model based inference process, please refer to document GEMX based Keras MLP Acceleration.