19 …d_device(const T *sT, T rU[DIM_U][NUM_COMP][rU_SIZE], T rV[DIM_V][NUM_COMP][rV_SIZE], const int tx,  in magma_interp_2d_device()  argument
34     if (tx < P) {  in magma_interp_2d_device()
35       const int batchid = tx;  in magma_interp_2d_device()
49     if (tx < Q) {  in magma_interp_2d_device()
56           rTmp += sTmp(tx, i, sld) * sT(i, j);  in magma_interp_2d_device()
71   const int tx      = threadIdx.x;  in __launch_bounds__()
92     read_T_notrans_gm2sm<BASIS_P, BASIS_Q>(tx, dT, sT);  in __launch_bounds__()
96   read_U_2d<CeedScalar, BASIS_P, 1, BASIS_NUM_COMP, BASIS_P, 0>(dU, cstrdU, rU, sTmp, tx);  in __launch_bounds__()
99 …<CeedScalar, 1, 1, BASIS_NUM_COMP, BASIS_P, BASIS_Q, BASIS_P, BASIS_Q>(sT, rU, rV, tx, rTmp, sTmp);  in __launch_bounds__()
103   write_V_2d<CeedScalar, BASIS_Q, 1, BASIS_NUM_COMP, BASIS_Q, 0>(dV, cstrdV, rV, tx);  in __launch_bounds__()
112   const int tx      = threadIdx.x;  in __launch_bounds__()
133     read_T_trans_gm2sm<BASIS_Q, BASIS_P>(tx, dT, sT);  in __launch_bounds__()
137   read_U_2d<CeedScalar, BASIS_Q, 1, BASIS_NUM_COMP, BASIS_Q, 0>(dU, cstrdU, rU, sTmp, tx);  in __launch_bounds__()
140 …<CeedScalar, 1, 1, BASIS_NUM_COMP, BASIS_Q, BASIS_P, BASIS_Q, BASIS_P>(sT, rU, rV, tx, rTmp, sTmp);  in __launch_bounds__()
144   write_V_2d<CeedScalar, BASIS_P, 1, BASIS_NUM_COMP, BASIS_P, 0>(dV, cstrdV, rV, tx);  in __launch_bounds__()
153   const int tx      = threadIdx.x;  in __launch_bounds__()
174     read_T_trans_gm2sm<BASIS_Q, BASIS_P>(tx, dT, sT);  in __launch_bounds__()
178   read_U_2d<CeedScalar, BASIS_Q, 1, BASIS_NUM_COMP, BASIS_Q, 0>(dU, cstrdU, rU, sTmp, tx);  in __launch_bounds__()
181 …<CeedScalar, 1, 1, BASIS_NUM_COMP, BASIS_Q, BASIS_P, BASIS_Q, BASIS_P>(sT, rU, rV, tx, rTmp, sTmp);  in __launch_bounds__()
185   sum_V_2d<CeedScalar, BASIS_P, 1, BASIS_NUM_COMP, BASIS_P, 0>(dV, cstrdV, rV, tx);  in __launch_bounds__()