FPGAs, or field-programmable gate arrays, are semiconductor devices that can be reprogrammed to specific functionality or application requirements after manufacture.
“The convergence of several needs and technology progress makes FPGA attractive for HPC systems as an alternative to other accelerator technologies,” said Franck Cappello, a senior scientist in the Mathematics and Computer Science Division at Argonne and a coorganizer of the workshop.
FPGAs and reconfigurable computing in general offer an attractive avenue to address the important new needs of scientific computing and data analytics: significant increase in performance per watt and faster processing of data analysis. Advanced FPGAs combine (1) new system-on-chip devices featuring multicore CPUs, FPGAs, thousands of floating-point data signal processing blocks, and tens of high-bandwidth transceivers and (2) advanced OpenCL compiler technologies capable of targeting heterogeneous systems. These, combined with the push at application level toward accelerator-oriented parallel programming with APIs (e.g., OpenMP, OpenCL), open up FPGA-based solutions for serious exploration in scientific simulations and data analytics.
The workshop will feature presentations from experts in the field and vendors (Xilinx, Intel/Altera) and will include ample time for discussion and special sessions for exploring potential collaborations.
“Our aim is to provide an opportunity not only to examine the state of the art and emerging trends in FPGA technologies but also to better understand how to address the challenges raised by reconfigurable technologies,” said Cappello,
For further information, see the website: