42 …TensorContract_Avx_Blocked(CeedTensorContract contract, CeedInt A, CeedInt B, CeedInt C, CeedInt J,  in CeedTensorContract_Avx_Blocked()  argument
49     t_stride_1 = J;  in CeedTensorContract_Avx_Blocked()
54     for (CeedInt j = 0; j < (J / JJ) * JJ; j += JJ) {  in CeedTensorContract_Avx_Blocked()
58 … for (CeedInt cc = 0; cc < CC / 4; cc++) vv[jj][cc] = loadu(&v[(a * J + j + jj) * C + c + cc * 4]);  in CeedTensorContract_Avx_Blocked()
69 … for (CeedInt cc = 0; cc < CC / 4; cc++) storeu(&v[(a * J + j + jj) * C + c + cc * 4], vv[jj][cc]);  in CeedTensorContract_Avx_Blocked()
74     const CeedInt j = (J / JJ) * JJ;  in CeedTensorContract_Avx_Blocked()
76     if (j < J) {  in CeedTensorContract_Avx_Blocked()
80         for (CeedInt jj = 0; jj < J - j; jj++) {  in CeedTensorContract_Avx_Blocked()
81 … for (CeedInt cc = 0; cc < CC / 4; cc++) vv[jj][cc] = loadu(&v[(a * J + j + jj) * C + c + cc * 4]);  in CeedTensorContract_Avx_Blocked()
84           for (CeedInt jj = 0; jj < J - j; jj++) {  // doesn't unroll  in CeedTensorContract_Avx_Blocked()
92         for (CeedInt jj = 0; jj < J - j; jj++) {  in CeedTensorContract_Avx_Blocked()
93 … for (CeedInt cc = 0; cc < CC / 4; cc++) storeu(&v[(a * J + j + jj) * C + c + cc * 4], vv[jj][cc]);  in CeedTensorContract_Avx_Blocked()
104 …nsorContract_Avx_Remainder(CeedTensorContract contract, CeedInt A, CeedInt B, CeedInt C, CeedInt J,  in CeedTensorContract_Avx_Remainder()  argument
111     t_stride_1 = J;  in CeedTensorContract_Avx_Remainder()
114   const CeedInt J_break = J % JJ ? (J / JJ) * JJ : (J / JJ - 1) * JJ;  in CeedTensorContract_Avx_Remainder()
123         for (CeedInt jj = 0; jj < JJ; jj++) vv[jj] = loadu(&v[(a * J + j + jj) * C + c]);  in CeedTensorContract_Avx_Remainder()
135         for (CeedInt jj = 0; jj < JJ; jj++) storeu(&v[(a * J + j + jj) * C + c], vv[jj]);  in CeedTensorContract_Avx_Remainder()
139     for (CeedInt j = J_break; j < J; j++) {  in CeedTensorContract_Avx_Remainder()
143 … for (CeedInt c = (C / CC) * CC; c < C; c++) v[(a * J + j) * C + c] += tq * u[(a * B + b) * C + c];  in CeedTensorContract_Avx_Remainder()
153 …ingle(CeedTensorContract contract, CeedInt A, CeedInt B, CeedInt C, CeedInt J, const CeedScalar *r…  in CeedTensorContract_Avx_Single()  argument
160     t_stride_1 = J;  in CeedTensorContract_Avx_Single()
165     for (CeedInt j = 0; j < (J / JJ) * JJ; j += JJ) {  in CeedTensorContract_Avx_Single()
169         for (CeedInt jj = 0; jj < JJ / 4; jj++) vv[aa][jj] = loadu(&v[(a + aa) * J + j + jj * 4]);  in CeedTensorContract_Avx_Single()
182         for (CeedInt jj = 0; jj < JJ / 4; jj++) storeu(&v[(a + aa) * J + j + jj * 4], vv[aa][jj]);  in CeedTensorContract_Avx_Single()
189   for (CeedInt j = 0; j < (J / JJ) * JJ; j += JJ) {  in CeedTensorContract_Avx_Single()
193       for (CeedInt jj = 0; jj < JJ / 4; jj++) vv[aa][jj] = loadu(&v[(a + aa) * J + j + jj * 4]);  in CeedTensorContract_Avx_Single()
206       for (CeedInt jj = 0; jj < JJ / 4; jj++) storeu(&v[(a + aa) * J + j + jj * 4], vv[aa][jj]);  in CeedTensorContract_Avx_Single()
213   for (CeedInt j = (J / JJ) * JJ; j < J; j += 4) {  in CeedTensorContract_Avx_Single()
218       for (CeedInt aa = 0; aa < AA; aa++) vv[aa] = loadu(&v[(a + aa) * J + j]);  in CeedTensorContract_Avx_Single()
222         if (J - j == 1) {  in CeedTensorContract_Avx_Single()
224         } else if (J - j == 2) {  in CeedTensorContract_Avx_Single()
226         } else if (J - 3 == j) {  in CeedTensorContract_Avx_Single()
237       for (CeedInt aa = 0; aa < AA; aa++) storeu(&v[(a + aa) * J + j], vv[aa]);  in CeedTensorContract_Avx_Single()
242     for (CeedInt j = (J / JJ) * JJ; j < J; j++) {  in CeedTensorContract_Avx_Single()
245       for (CeedInt a = A_break; a < A; a++) v[a * J + j] += tq * u[a * B + b];  in CeedTensorContract_Avx_Single()
254 …d_4_8(CeedTensorContract contract, CeedInt A, CeedInt B, CeedInt C, CeedInt J, const CeedScalar *r…  in CeedTensorContract_Avx_Blocked_4_8()  argument
256   return CeedTensorContract_Avx_Blocked(contract, A, B, C, J, t, t_mode, add, u, v, 4, 8);  in CeedTensorContract_Avx_Blocked_4_8()
258 …r_8_8(CeedTensorContract contract, CeedInt A, CeedInt B, CeedInt C, CeedInt J, const CeedScalar *r…  in CeedTensorContract_Avx_Remainder_8_8()  argument
260   return CeedTensorContract_Avx_Remainder(contract, A, B, C, J, t, t_mode, add, u, v, 8, 8);  in CeedTensorContract_Avx_Remainder_8_8()
262 …e_4_8(CeedTensorContract contract, CeedInt A, CeedInt B, CeedInt C, CeedInt J, const CeedScalar *r…  in CeedTensorContract_Avx_Single_4_8()  argument
264   return CeedTensorContract_Avx_Single(contract, A, B, C, J, t, t_mode, add, u, v, 4, 8);  in CeedTensorContract_Avx_Single_4_8()
270 …y_Avx(CeedTensorContract contract, CeedInt A, CeedInt B, CeedInt C, CeedInt J, const CeedScalar *r…  in CeedTensorContractApply_Avx()  argument
275     for (CeedInt q = 0; q < A * J * C; q++) v[q] = (CeedScalar)0.0;  in CeedTensorContractApply_Avx()
280     CeedTensorContract_Avx_Single_4_8(contract, A, B, C, J, t, t_mode, true, u, v);  in CeedTensorContractApply_Avx()
283 …if (C >= blk_size) CeedTensorContract_Avx_Blocked_4_8(contract, A, B, C, J, t, t_mode, true, u, v);  in CeedTensorContractApply_Avx()
285 …if (C % blk_size) CeedTensorContract_Avx_Remainder_8_8(contract, A, B, C, J, t, t_mode, true, u, v…  in CeedTensorContractApply_Avx()