44 …        const CeedScalar *restrict u, CeedScalar *restrict v, const CeedInt JJ, const CeedInt CC) {  in CeedTensorContract_Avx_Blocked()  argument
55       for (CeedInt c = 0; c < (C / CC) * CC; c += CC) {  in CeedTensorContract_Avx_Blocked()
56         rtype vv[JJ][CC / 4];  // Output tile to be held in registers  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()  local
63             for (CeedInt cc = 0; cc < CC / 4; cc++) {  // unroll  in CeedTensorContract_Avx_Blocked()  local
64               fmadd(vv[jj][cc], tqv, loadu(&u[(a * B + b) * 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()  local
77       for (CeedInt c = 0; c < (C / CC) * CC; c += CC) {  in CeedTensorContract_Avx_Blocked()
78         rtype vv[JJ][CC / 4];  // Output tile to be held in registers  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()  local
87             for (CeedInt cc = 0; cc < CC / 4; cc++) {  // unroll  in CeedTensorContract_Avx_Blocked()  local
88               fmadd(vv[jj][cc], tqv, loadu(&u[(a * B + b) * C + c + cc * 4]));  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()  local
106 …        const CeedScalar *restrict u, CeedScalar *restrict v, const CeedInt JJ, const CeedInt CC) {  in CeedTensorContract_Avx_Remainder()  argument
118     for (CeedInt c = (C / CC) * CC; c < C; c += 4) {  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()