128         CeedInt elems_per_block = 64 * thread_1d > 256 ? 256 / thread_1d : 64;  in CeedBasisApplyTensorCore_Hip_shared()  local
141         const CeedInt elems_per_block = CeedIntMax(block_size / (thread_1d * thread_1d), 1);  in CeedBasisApplyTensorCore_Hip_shared()  local
152         const CeedInt elems_per_block = CeedIntMax(block_size / (thread_1d * thread_1d), 1);  in CeedBasisApplyTensorCore_Hip_shared()  local
180         CeedInt elems_per_block = 64 * thread_1d > 256 ? 256 / thread_1d : 64;  in CeedBasisApplyTensorCore_Hip_shared()  local
193         const CeedInt elems_per_block = CeedIntMax(block_size / (thread_1d * thread_1d), 1);  in CeedBasisApplyTensorCore_Hip_shared()  local
204         const CeedInt elems_per_block = CeedIntMax(block_size / (thread_1d * thread_1d), 1);  in CeedBasisApplyTensorCore_Hip_shared()  local
226         const CeedInt elems_per_block = opt_elems > 0 ? opt_elems : 1;  in CeedBasisApplyTensorCore_Hip_shared()  local
232         const CeedInt elems_per_block = opt_elems > 0 ? opt_elems : 1;  in CeedBasisApplyTensorCore_Hip_shared()  local
238         const CeedInt elems_per_block = opt_elems > 0 ? opt_elems : 1;  in CeedBasisApplyTensorCore_Hip_shared()  local
391         CeedInt elems_per_block = 64 * thread_1d > 256 ? 256 / thread_1d : 64;  in CeedBasisApplyAtPointsCore_Hip_shared()  local
404         const CeedInt elems_per_block = CeedIntMax(block_size / (thread_1d * thread_1d), 1);  in CeedBasisApplyAtPointsCore_Hip_shared()  local
416         const CeedInt elems_per_block = 1;  in CeedBasisApplyAtPointsCore_Hip_shared()  local
439         CeedInt elems_per_block = 64 * thread_1d > 256 ? 256 / thread_1d : 64;  in CeedBasisApplyAtPointsCore_Hip_shared()  local
452         const CeedInt elems_per_block = CeedIntMax(block_size / (thread_1d * thread_1d), 1);  in CeedBasisApplyAtPointsCore_Hip_shared()  local
464         const CeedInt elems_per_block = 1;  in CeedBasisApplyAtPointsCore_Hip_shared()  local
547         CeedInt elems_per_block = 64 * thread > 256 ? 256 / thread : 64;  in CeedBasisApplyNonTensorCore_Hip_shared()  local
570         CeedInt elems_per_block = 64 * thread > 256 ? 256 / thread : 64;  in CeedBasisApplyNonTensorCore_Hip_shared()  local
593         CeedInt elems_per_block = 64 * thread > 256 ? 256 / thread : 64;  in CeedBasisApplyNonTensorCore_Hip_shared()  local