Extending Parallel Python Computing to FPGAs

Extending Parallel Python Computing to FPGAs

January 4, 2022
Behzad Nasre

Today I’m happy to announce our collaboration with Xilinx, Inc., including their taking an investment in Bodo. This will be meaningful to more than just Xilinx and Bodo customers... It signifies another stage of our democratizing access to large-scale parallel computing using Python.

Late last year we showed that for large-scale analytics computing, Bodo’s parallel approach was far faster, more efficient, and scaled better than Spark or Dask, yet was based on using standard CPU clusters.

Today we’ve taken another step in our strategy, beginning work with Xilinx to develop the same highly-efficient and scalable parallel computing for media on FPGA-enabled clusters - again for use with Python.

Media developers today approach performance and scale by developing with underlying hardware in mind. This requires special skills, languages, code rewrites and tuning. Until now, high-level languages such as Python were not optimized for this type of computation. In fact, media, tabular, and semi-structured data processing each use different programming approaches and differing optimal hardware.

Our collaboration will allow media developer customers using Xilinx to develop in familiar languages like Python, yet take advantage of Bodo’s portable parallel computing architecture to simplify scaling across 10,000’s of cores.

Looking forward, Bodo plans to continue our direction: To provide developers with technologies that not only exploit the efficiencies of parallel computing, but also use standard high-level languages, while optimizing for the underlying hardware.

By using this website, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Policy for more information.