1 /* 2 Copyright (c) 2011, Intel Corporation. All rights reserved. 3 4 Redistribution and use in source and binary forms, with or without modification, 5 are permitted provided that the following conditions are met: 6 7 * Redistributions of source code must retain the above copyright notice, this 8 list of conditions and the following disclaimer. 9 * Redistributions in binary form must reproduce the above copyright notice, 10 this list of conditions and the following disclaimer in the documentation 11 and/or other materials provided with the distribution. 12 * Neither the name of Intel Corporation nor the names of its contributors may 13 be used to endorse or promote products derived from this software without 14 specific prior written permission. 15 16 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND 17 ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED 18 WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE 19 DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR 20 ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES 21 (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 22 LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON 23 ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 24 (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 25 SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 26 27 ******************************************************************************** 28 * Content : Eigen bindings to Intel(R) MKL 29 * General matrix-matrix product functionality based on ?GEMM. 30 ******************************************************************************** 31 */ 32 33 #ifndef EIGEN_GENERAL_MATRIX_MATRIX_MKL_H 34 #define EIGEN_GENERAL_MATRIX_MATRIX_MKL_H 35 36 namespace Eigen { 37 38 namespace internal { 39 40 /********************************************************************** 41 * This file implements general matrix-matrix multiplication using BLAS 42 * gemm function via partial specialization of 43 * general_matrix_matrix_product::run(..) method for float, double, 44 * std::complex<float> and std::complex<double> types 45 **********************************************************************/ 46 47 // gemm specialization 48 49 #define GEMM_SPECIALIZATION(EIGTYPE, EIGPREFIX, MKLTYPE, MKLPREFIX) \ 50 template< \ 51 typename Index, \ 52 int LhsStorageOrder, bool ConjugateLhs, \ 53 int RhsStorageOrder, bool ConjugateRhs> \ 54 struct general_matrix_matrix_product<Index,EIGTYPE,LhsStorageOrder,ConjugateLhs,EIGTYPE,RhsStorageOrder,ConjugateRhs,ColMajor> \ 55 { \ 56 static void run(Index rows, Index cols, Index depth, \ 57 const EIGTYPE* _lhs, Index lhsStride, \ 58 const EIGTYPE* _rhs, Index rhsStride, \ 59 EIGTYPE* res, Index resStride, \ 60 EIGTYPE alpha, \ 61 level3_blocking<EIGTYPE, EIGTYPE>& /*blocking*/, \ 62 GemmParallelInfo<Index>* /*info = 0*/) \ 63 { \ 64 using std::conj; \ 65 \ 66 char transa, transb; \ 67 MKL_INT m, n, k, lda, ldb, ldc; \ 68 const EIGTYPE *a, *b; \ 69 MKLTYPE alpha_, beta_; \ 70 MatrixX##EIGPREFIX a_tmp, b_tmp; \ 71 EIGTYPE myone(1);\ 72 \ 73 /* Set transpose options */ \ 74 transa = (LhsStorageOrder==RowMajor) ? ((ConjugateLhs) ? 'C' : 'T') : 'N'; \ 75 transb = (RhsStorageOrder==RowMajor) ? ((ConjugateRhs) ? 'C' : 'T') : 'N'; \ 76 \ 77 /* Set m, n, k */ \ 78 m = (MKL_INT)rows; \ 79 n = (MKL_INT)cols; \ 80 k = (MKL_INT)depth; \ 81 \ 82 /* Set alpha_ & beta_ */ \ 83 assign_scalar_eig2mkl(alpha_, alpha); \ 84 assign_scalar_eig2mkl(beta_, myone); \ 85 \ 86 /* Set lda, ldb, ldc */ \ 87 lda = (MKL_INT)lhsStride; \ 88 ldb = (MKL_INT)rhsStride; \ 89 ldc = (MKL_INT)resStride; \ 90 \ 91 /* Set a, b, c */ \ 92 if ((LhsStorageOrder==ColMajor) && (ConjugateLhs)) { \ 93 Map<const MatrixX##EIGPREFIX, 0, OuterStride<> > lhs(_lhs,m,k,OuterStride<>(lhsStride)); \ 94 a_tmp = lhs.conjugate(); \ 95 a = a_tmp.data(); \ 96 lda = a_tmp.outerStride(); \ 97 } else a = _lhs; \ 98 \ 99 if ((RhsStorageOrder==ColMajor) && (ConjugateRhs)) { \ 100 Map<const MatrixX##EIGPREFIX, 0, OuterStride<> > rhs(_rhs,k,n,OuterStride<>(rhsStride)); \ 101 b_tmp = rhs.conjugate(); \ 102 b = b_tmp.data(); \ 103 ldb = b_tmp.outerStride(); \ 104 } else b = _rhs; \ 105 \ 106 MKLPREFIX##gemm(&transa, &transb, &m, &n, &k, &alpha_, (const MKLTYPE*)a, &lda, (const MKLTYPE*)b, &ldb, &beta_, (MKLTYPE*)res, &ldc); \ 107 }}; 108 109 GEMM_SPECIALIZATION(double, d, double, d) 110 GEMM_SPECIALIZATION(float, f, float, s) 111 GEMM_SPECIALIZATION(dcomplex, cd, MKL_Complex16, z) 112 GEMM_SPECIALIZATION(scomplex, cf, MKL_Complex8, c) 113 114 } // end namespase internal 115 116 } // end namespace Eigen 117 118 #endif // EIGEN_GENERAL_MATRIX_MATRIX_MKL_H 119