Simon Brown@sjb3dFinally got streaming/compaction model to be faster than (properly pipelined) megakernels in #CUDA. 2300 kernels in 550ms!9:00 AM · Jun 18, 2013·Twitter Web Client3 Likes
