The Bambu HLS flow is available on the
dev
branch and is planned for the next ESP release. Coming soon
Platform support: like ESP’s other HLS flows, the Bambu flow targets AMD/Xilinx FPGA technologies. Virtex-7 boards are mapped automatically; other AMD/Xilinx technologies need an explicit Bambu device setting.
What you will learn
- Generate a Bambu HLS accelerator skeleton with ESP’s accelerator generator
- C/C++ dataflow kernel with load, compute and store stages
- C++ testbench and ESP socket wrapper
- bare-metal, device-driver and user-space application skeletons
- Synthesize the accelerator with Bambu and install it into ESP
- Verify the accelerator natively and with Bambu RTL co-simulation
- Integrate the accelerator into an ESP SoC, simulate it and test it on FPGA
What you will need
- Prerequisites
- Set up ESP, including the optional PandA Bambu installation
- Design a single-core SoC
- Equipment
- A supported AMD/Xilinx FPGA board; the example was validated on a VC707
What you can contribute
The ESP team welcomes external contributions and collaborations, including:
- Accelerator designs for a wide range of application domains
- Chunking, batching and additional data widths for the Bambu flow
- Bambu device mappings for more FPGA technologies
See the contributing guidelines.