1 /* 2 3 BLIS 4 An object-based framework for developing high-performance BLAS-like 5 libraries. 6 7 Copyright (C) 2014, The University of Texas at Austin 8 9 Redistribution and use in source and binary forms, with or without 10 modification, are permitted provided that the following conditions are 11 met: 12 - Redistributions of source code must retain the above copyright 13 notice, this list of conditions and the following disclaimer. 14 - Redistributions in binary form must reproduce the above copyright 15 notice, this list of conditions and the following disclaimer in the 16 documentation and/or other materials provided with the distribution. 17 - Neither the name(s) of the copyright holder(s) nor the names of its 18 contributors may be used to endorse or promote products derived 19 from this software without specific prior written permission. 20 21 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 22 "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 23 LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 24 A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT 25 HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 26 SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 27 LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 28 DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 29 THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 30 (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE 31 OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 32 33 */ 34 35 #include "blis.h" 36 37 38 #undef GENTFUNC 39 #define GENTFUNC( ctype, ch, varname ) \ 40 \ 41 void PASTEMAC(ch,varname) \ 42 ( \ 43 conj_t conjat, \ 44 conj_t conja, \ 45 conj_t conjw, \ 46 conj_t conjx, \ 47 dim_t m, \ 48 dim_t b_n, \ 49 ctype* restrict alpha, \ 50 ctype* restrict a, inc_t inca, inc_t lda, \ 51 ctype* restrict w, inc_t incw, \ 52 ctype* restrict x, inc_t incx, \ 53 ctype* restrict beta, \ 54 ctype* restrict y, inc_t incy, \ 55 ctype* restrict z, inc_t incz, \ 56 cntx_t* cntx \ 57 ) \ 58 { \ 59 ctype* a1; \ 60 ctype* chi1; \ 61 ctype* w1; \ 62 ctype* psi1; \ 63 ctype* z1; \ 64 ctype conjx_chi1; \ 65 ctype alpha_chi1; \ 66 dim_t i; \ 67 \ 68 /* Query the context for the kernel function pointer. */ \ 69 const num_t dt = PASTEMAC(ch,type); \ 70 PASTECH(ch,dotxv_ft) kfp_dv = bli_cntx_get_l1v_ker_dt( dt, BLIS_DOTXV_KER, cntx ); \ 71 PASTECH(ch,axpyv_ft) kfp_av = bli_cntx_get_l1v_ker_dt( dt, BLIS_AXPYV_KER, cntx ); \ 72 \ 73 /* A is m x n. */ \ 74 /* y = beta * y + alpha * A^T w; */ \ 75 /* z = z + alpha * A x; */ \ 76 for ( i = 0; i < b_n; ++i ) \ 77 { \ 78 a1 = a + (0 )*inca + (i )*lda; \ 79 w1 = w + (0 )*incw; \ 80 psi1 = y + (i )*incy; \ 81 \ 82 kfp_dv \ 83 ( \ 84 conjat, \ 85 conjw, \ 86 m, \ 87 alpha, \ 88 a1, inca, \ 89 w1, incw, \ 90 beta, \ 91 psi1, \ 92 cntx \ 93 ); \ 94 } \ 95 \ 96 for ( i = 0; i < b_n; ++i ) \ 97 { \ 98 a1 = a + (0 )*inca + (i )*lda; \ 99 chi1 = x + (i )*incx; \ 100 z1 = z + (0 )*incz; \ 101 \ 102 PASTEMAC(ch,copycjs)( conjx, *chi1, conjx_chi1 ); \ 103 PASTEMAC(ch,scal2s)( *alpha, conjx_chi1, alpha_chi1 ); \ 104 \ 105 kfp_av \ 106 ( \ 107 conja, \ 108 m, \ 109 &alpha_chi1, \ 110 a1, inca, \ 111 z1, incz, \ 112 cntx \ 113 ); \ 114 } \ 115 } 116 117 INSERT_GENTFUNC_BASIC0( dotxaxpyf_ref_var1 ) 118 119