Skip to content

Hotfix: supporting non-scalar fold operations and mo-WLs within CUDA kernels

Sven-Bodo Scholz requested to merge sbs/sac2c:hotfix-mowl-cuknl into develop

There were two distinct problems.

  1. the return values of mows were lifted in the wrong order in cuknl.
  2. the initialisation of fold-wls inside of CUDA kernels went wrong

Both of these issues are fixed here and one test for each has been added.

Merge request reports