Comparing Energy Efficiency of CPU, GPU and FPGA Implementations for Vision Kernels — mentioned by cariboulabs

Sun, 05 Dec 2021 07:01:26 GMT

power efficiency was proven to be in the same orders of magnitude as MPU / GPU. see e.g. https://arxiv.org/pdf/1906.11879.pdf. The ability to mutate its firmware gives another crucial edge. And my point is that FPGAs have evolved is small paces until today. While dedicated silicon has been optimized to the bones. Given the same dedication of companies to FPGAs - isn't it going to evolve into a better, cheaper device? More interesting info: see RiscV soft cores, SDRs, MCUs with internal logic fabric.

Paper and discussion history. A mention is not an endorsement.

Comparing Energy Efficiency of CPU, GPU and FPGA Implementations for Vision Kernels · HackerPapers discussions

Comparing Energy Efficiency of CPU, GPU and FPGA Implementations for Vision Kernels — mentioned by cariboulabs