This survey paper published in ACM Computing Surveys 2015 compares GPU with FPGA and CPU on energy efficiency metric. Most papers reviewed in the surv ey report that FPGA is more energy efficient than GPU, which, in turn, is m ore energy efficient than CPU.
For one specific use, which is not the use that either GPUs or FPGAs were designed.
(Of course, it is, theoretically, the use for which CPUs were designed, and that's the one that sucks the worst -- go figure).
I did not read the paper. Did it say anything about the ease of programming for GPUs vs. FPGA in the applications they were looking at? In my (limited) experience with DSP vs. FPGA tradeoffs for video application, the ultimate performance limit was memory bandwidth -- did this paper address inter-processor communications and memory issues, or was it just doing some artificial test like Dhrystones or MFLOP/Joule or something?
Have something to add? Share your thoughts — no account required.
Ask the community — no account required