Advanced algorithmic techniques for GPUs
University of Illinois, , Prof. Wen-mei Hwu
Added to favorite list
Updated On 02 Feb, 19
4.1 ( 11 )
Often domain problems have inherent parallelism that needs to be recognized. The most efficient implementation that exploits the problemís parallelism may be non-intuitive. For example, two alternative thread arrangements that appear in electrostatics calculations have, respectively, scatter and gather memory access behavior. The first is more intuitive, but the second is much more efficient on the GPU architecture.
Sep 12, 2018
Excellent course helped me understand topic that i couldn't while attendinfg my college.
March 29, 2019
Great course. Thank you very much.