Commit c095ee4f authored by Stephan Seitz's avatar Stephan Seitz
Browse files

Bugfix: Align calculation of number of ghost layers on GPU with CPU version

For the calculation of the number of ghostlayers only relative accesses
should be considered like on the CPU version
parent 0ff425f3
Pipeline #18497 passed with stage
in 5 minutes and 30 seconds
...@@ -27,6 +27,7 @@ def create_cuda_kernel(assignments, ...@@ -27,6 +27,7 @@ def create_cuda_kernel(assignments,
num_buffer_accesses = 0 num_buffer_accesses = 0
for eq in assignments: for eq in assignments:
field_accesses.update(eq.atoms(Field.Access)) field_accesses.update(eq.atoms(Field.Access))
field_accesses = {e for e in field_accesses if not e.is_absolute_access}
num_buffer_accesses += sum(1 for access in eq.atoms(Field.Access) if FieldType.is_buffer(access.field)) num_buffer_accesses += sum(1 for access in eq.atoms(Field.Access) if FieldType.is_buffer(access.field))
common_shape = get_common_shape(fields_without_buffers) common_shape = get_common_shape(fields_without_buffers)
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment