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 #undef GENTFUNC 38 #define GENTFUNC( ctype, ch, varname ) \ 39 \ 40 void PASTEMAC(ch,varname) \ 41 ( \ 42 trans_t transa, \ 43 conj_t conjx, \ 44 dim_t m, \ 45 dim_t n, \ 46 ctype* alpha, \ 47 ctype* a, inc_t rs_a, inc_t cs_a, \ 48 ctype* x, inc_t incx, \ 49 ctype* beta, \ 50 ctype* y, inc_t incy, \ 51 cntx_t* cntx \ 52 ) \ 53 { \ 54 const num_t dt = PASTEMAC(ch,type); \ 55 \ 56 ctype* zero = PASTEMAC(ch,0); \ 57 ctype* A1; \ 58 ctype* x1; \ 59 ctype* y1; \ 60 dim_t i; \ 61 dim_t b_fuse, f; \ 62 dim_t n_elem, n_iter; \ 63 inc_t rs_at, cs_at; \ 64 conj_t conja; \ 65 \ 66 bli_set_dims_incs_with_trans( transa, \ 67 m, n, rs_a, cs_a, \ 68 &n_elem, &n_iter, &rs_at, &cs_at ); \ 69 \ 70 conja = bli_extract_conj( transa ); \ 71 \ 72 /* If beta is zero, use setv. Otherwise, scale by beta. */ \ 73 if ( PASTEMAC(ch,eq0)( *beta ) ) \ 74 { \ 75 /* y = 0; */ \ 76 PASTEMAC2(ch,setv,BLIS_TAPI_EX_SUF) \ 77 ( \ 78 BLIS_NO_CONJUGATE, \ 79 n_elem, \ 80 zero, \ 81 y, incy, \ 82 cntx, \ 83 NULL \ 84 ); \ 85 } \ 86 else \ 87 { \ 88 /* y = beta * y; */ \ 89 PASTEMAC2(ch,scalv,BLIS_TAPI_EX_SUF) \ 90 ( \ 91 BLIS_NO_CONJUGATE, \ 92 n_elem, \ 93 beta, \ 94 y, incy, \ 95 cntx, \ 96 NULL \ 97 ); \ 98 } \ 99 \ 100 PASTECH(ch,axpyf_ker_ft) kfp_af; \ 101 \ 102 /* Query the context for the kernel function pointer and fusing factor. */ \ 103 kfp_af = bli_cntx_get_l1f_ker_dt( dt, BLIS_AXPYF_KER, cntx ); \ 104 b_fuse = bli_cntx_get_blksz_def_dt( dt, BLIS_AF, cntx ); \ 105 \ 106 for ( i = 0; i < n_iter; i += f ) \ 107 { \ 108 f = bli_determine_blocksize_dim_f( i, n_iter, b_fuse ); \ 109 \ 110 A1 = a + (0 )*rs_at + (i )*cs_at; \ 111 x1 = x + (i )*incx; \ 112 y1 = y + (0 )*incy; \ 113 \ 114 /* y = y + alpha * A1 * x1; */ \ 115 kfp_af \ 116 ( \ 117 conja, \ 118 conjx, \ 119 n_elem, \ 120 f, \ 121 alpha, \ 122 A1, rs_at, cs_at, \ 123 x1, incx, \ 124 y1, incy, \ 125 cntx \ 126 ); \ 127 } \ 128 } 129 130 INSERT_GENTFUNC_BASIC0( gemv_unf_var2 ) 131 132