Hi,
Do we have to allocate memory as unified for the data to use within a grid_group?
Regards,
Buket
Hi,
Do we have to allocate memory as unified for the data to use within a grid_group?
Regards,
Buket
There aren’t any memory allocation requirements that are specific to cooperative groups. Device memory allocated via either cudaMalloc
or cudaMallocManaged
should be accessible to every member of a grid_group
.
Thanks a lot.
Regards,
Buket