44 …        const CeedScalar *restrict u, CeedScalar *restrict v, const CeedInt JJ, const CeedInt CC) {  in CeedTensorContract_Avx_Blocked()  argument
54     for (CeedInt j = 0; j < (J / JJ) * JJ; j += JJ) {  in CeedTensorContract_Avx_Blocked()
56         rtype vv[JJ][CC / 4];  // Output tile to be held in registers  in CeedTensorContract_Avx_Blocked()
57         for (CeedInt jj = 0; jj < JJ; jj++) {  in CeedTensorContract_Avx_Blocked()
61           for (CeedInt jj = 0; jj < JJ; jj++) {  // unroll  in CeedTensorContract_Avx_Blocked()
68         for (CeedInt jj = 0; jj < JJ; jj++) {  in CeedTensorContract_Avx_Blocked()
74     const CeedInt j = (J / JJ) * JJ;  in CeedTensorContract_Avx_Blocked()
78         rtype vv[JJ][CC / 4];  // Output tile to be held in registers  in CeedTensorContract_Avx_Blocked()
106 …        const CeedScalar *restrict u, CeedScalar *restrict v, const CeedInt JJ, const CeedInt CC) {  in CeedTensorContract_Avx_Remainder()  argument
114   const CeedInt J_break = J % JJ ? (J / JJ) * JJ : (J / JJ - 1) * JJ;  in CeedTensorContract_Avx_Remainder()
120       for (CeedInt j = 0; j < J_break; j += JJ) {  in CeedTensorContract_Avx_Remainder()
121         rtype vv[JJ];  // Output tile to be held in registers  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()
131           for (CeedInt jj = 0; jj < JJ; jj++) {  // unroll  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()
155                                                 const CeedInt AA, const CeedInt JJ) {  in CeedTensorContract_Avx_Single()  argument
165     for (CeedInt j = 0; j < (J / JJ) * JJ; j += JJ) {  in CeedTensorContract_Avx_Single()
166       rtype vv[AA][JJ / 4];  // Output tile to be held in registers  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()
172         for (CeedInt jj = 0; jj < JJ / 4; jj++) {  // unroll  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()
190     rtype vv[AA][JJ / 4];  // Output tile to be held in registers  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()
196       for (CeedInt jj = 0; jj < JJ / 4; jj++) {  // unroll  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()
242     for (CeedInt j = (J / JJ) * JJ; j < J; j++) {  in CeedTensorContract_Avx_Single()