Webinar : First Dataflow Modeling of the DDFacet Imager for Acceleration Using FPGA Coprocessors

First Dataflow Modeling of the DDFacet Imager for Acceleration Using FPGA Coprocessors

This webinar presents the first dataflow modeling of the DDFacet imager, a cutting-edge radio astronomy imaging and deconvolution framework and how this dataflow model will be used to automatically evaluate the benefits of using FPGA coprocessors to accelerate DDFacet performance, addressing the growing computational demands of next-generation radio telescopes, real-time constraints and high-sensitivity radio imaging.

The use of these coprocessors must remain flexible in order to adapt the configuration of DDFacet to the specific needs of astronomers and to adapt to the available computing resources.

Key topics include:

• Introducing a dataflow model to analyze and optimize DDFacet’s processing pipeline, identifying bottlenecks and opportunities for hardware acceleration.
• Exploring how FPGAs can enhance computational efficiency, particularly for parallelizable tasks such as gridding/degridding and Fast Fourier Transforms