Copyright 2019 Xilinx, Inc.

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

This work culminated in HuDSoN. However, HuDSoN lives now in another repository: https://github.com/HansGiesen/hudson

This is the pragma tuner for Vivado HLS developed by Hans Giesen, an intern from the University of Pennsylvania, during his internship in San Jose, CA between 6/5/2019 and 8/23/2019. For an introduction to the tuner, please consult the presentation that is located in the Presentation directory.

The pragma tuner is based on OpenTuner. OpenTuner, which is located in the file OpenTuner/LICENSE.txt, is governed by a license that is provided in that same directory. All other content in the repository is governed by the license in HLS_tuner/LICENSE.

The tuner requires Python 2. Version 2.7.5 worked great for me. All required packages are described in requirements.txt. I installed them in a virtual environment using virtualenv version 15.1.0. That should be sufficient to run the tuner, which is located at HLS_tuner/hlstuner/hls_tuner.py.

Debugging is rather time-consuming if you have to wait for builds to complete. Hence, the tuner has an option --use-prebuilt to reuse an old build, which is located in the Prebuilt directory. The option will replace the latency with a number from a artificial design space, useful for debugging the surrogate models.

Graphs were plot using the Jupyter that is part of Anaconda 2019.03. The notebooks can be found in the Notebook. Directories are hardcoded, and some tuner sources have changed since I last ran some of the scripts, so you will likely encounter some issues when running this code.