opm-simulators icon indicating copy to clipboard operation
opm-simulators copied to clipboard
verified publisher icon OPM

Generalize thread block tuner

Open multitalentloes opened this issue 1 year ago • 0 comments

Adds a function that picks the best thread block size for a kernel. This avoid code duplication currently present in CuDILU and CuILU.

cuda events are also used to make the code more reliable.

multitalentloes avatar Jun 27 '24 14:06 multitalentloes