| /libCEED/include/ceed/jit-source/magma/ |
| H A D | magma-basis-grad-3d.h | 14 #define sTmp(i, j, ldw) sTmp[(j) * (ldw) + (i)] macro
63 T *sTmp = sW1 + batchid * (1 * Q);
69 sTmp(0, j, sld) = rTmp;
80 T *sTmp = sW1 + batchid * (Q * P); // sTmp is input
85 rTmp += sTmp(tx_, i, sld) * sT(i, j);
98 T *sTmp = sW2; // sTmp is input
102 rTmp += sTmp(tx, i, sld) * sT(i, j);
134 CeedScalar *sTmp = sTgrad + BASIS_P * BASIS_Q;
135 …sTmp += ty * (max(BASIS_P * BASIS_P * BASIS_P, (BASIS_P * BASIS_P * BASIS_Q) + (BASIS_P * BASIS_Q …
146 …read_U_3d<CeedScalar, BASIS_P, 1, BASIS_NUM_COMP, BASIS_P, 0>(dU + (0 * dstrdU), cstrdU, rU, sTmp,…
[all …]
|
| H A D | magma-basis-interp-3d.h | 14 #define sTmp(i, j, ldw) sTmp[(j) * (ldw) + (i)] macro
37 T *sTmp = swork + batchid * (1 * Q); in magma_interp_3d_device() local
43 sTmp(0, j, sld) = rTmp[0]; in magma_interp_3d_device()
53 T *sTmp = swork + batchid * (Q * P); // sTmp is input in magma_interp_3d_device() local
57 rTmp[j] += sTmp(tx_, i, sld) * sT(i, j); in magma_interp_3d_device()
68 T *sTmp = swork + batchid * (Q * Q); in magma_interp_3d_device() local
70 sTmp(tx_, j, sld) = rTmp[j]; in magma_interp_3d_device()
80 T *sTmp = swork; in magma_interp_3d_device() local
84 rTmp[0] += sTmp(tx, i, sld) * sT(i, j); in magma_interp_3d_device()
115 CeedScalar *sTmp = sT + BASIS_P * BASIS_Q; in __launch_bounds__() local
[all …]
|
| H A D | magma-basis-interp-2d.h | 14 #define sTmp(i, j, ldw) sTmp[(j) * (ldw) + (i)] macro
37 T *sTmp = swork + batchid * (1 * Q); in magma_interp_2d_device() local
43 sTmp(0, j, sld) = rTmp; in magma_interp_2d_device()
52 T *sTmp = swork + batchid * (Q * P); in magma_interp_2d_device() local
56 rTmp += sTmp(tx, i, sld) * sT(i, j); in magma_interp_2d_device()
87 CeedScalar *sTmp = sT + BASIS_P * BASIS_Q; in __launch_bounds__() local
88 sTmp += ty * (BASIS_P * BASIS_MAX_P_Q); 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__()
128 CeedScalar *sTmp = sT + BASIS_Q * BASIS_P; in __launch_bounds__() local
[all …]
|
| H A D | magma-basis-grad-2d.h | 14 #define sTmp(i, j, ldw) sTmp[(j) * (ldw) + (i)] macro
60 T *sTmp = swork + batchid * (1 * Q);
66 sTmp(0, j, sld) = rTmp;
76 T *sTmp = swork + batchid * (Q * P);
80 rTmp += sTmp(tx, i, sld) * sT(i, j);
112 CeedScalar *sTmp = sTgrad + BASIS_P * BASIS_Q;
113 sTmp += ty * (BASIS_P * BASIS_MAX_P_Q);
123 …read_U_2d<CeedScalar, BASIS_P, 1, BASIS_NUM_COMP, BASIS_P, 0>(dU + (0 * dstrdU), cstrdU, rU, sTmp,…
128 … sTmp);
135 … sTmp);
[all …]
|
| H A D | magma-common-tensor.h | 59 …ad_U_2d(const T *dU, const int compstride, T rU[DIM_U][NUM_COMP][rU_SIZE], T *sTmp, const int tx) { in read_U_2d() argument
74 sTmp[i * P + tx] = dU[comp * compstride + i * P + tx]; in read_U_2d()
81 rU[i_DIM][comp][i] = sTmp[tx * P + i]; in read_U_2d()
147 …ad_U_3d(const T *dU, const int compstride, T rU[DIM_U][NUM_COMP][rU_SIZE], T *sTmp, const int tx) { in read_U_3d() argument
162 sTmp[i * P * P + tx] = dU[comp * compstride + i * P * P + tx]; in read_U_3d()
169 rU[i_DIM][comp][i] = sTmp[tx * P + i]; in read_U_3d()
|