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, opname, arch, suf ) \ 40 \ 41 void PASTEMAC3(ch,opname,arch,suf) \ 42 ( \ 43 conj_t conjxt, \ 44 conj_t conjx, \ 45 conj_t conjy, \ 46 dim_t m, \ 47 ctype* restrict alpha, \ 48 ctype* restrict x, inc_t incx, \ 49 ctype* restrict y, inc_t incy, \ 50 ctype* restrict rho, \ 51 ctype* restrict z, inc_t incz, \ 52 cntx_t* restrict cntx \ 53 ) \ 54 { \ 55 if ( bli_zero_dim1( m ) ) return; \ 56 \ 57 if ( incz == 1 && incx == 1 && incy == 1 ) \ 58 { \ 59 if ( bli_is_noconj( conjx ) ) \ 60 { \ 61 conj_t conjxt_use = conjxt; \ 62 ctype dotxy; \ 63 \ 64 PASTEMAC(ch,set0s)( dotxy ); \ 65 \ 66 if ( bli_is_conj( conjy ) ) \ 67 bli_toggle_conj( &conjxt_use ); \ 68 \ 69 if ( bli_is_noconj( conjxt_use ) ) \ 70 { \ 71 PRAGMA_SIMD \ 72 for ( dim_t i = 0; i < m; ++i ) \ 73 { \ 74 PASTEMAC(ch,dots)( x[i], y[i], dotxy ); \ 75 PASTEMAC(ch,axpys)( *alpha, x[i], z[i] ); \ 76 } \ 77 } \ 78 else /* bli_is_conj( conjxt_use ) ) */ \ 79 { \ 80 PRAGMA_SIMD \ 81 for ( dim_t i = 0; i < m; ++i ) \ 82 { \ 83 PASTEMAC(ch,dotjs)( x[i], y[i], dotxy ); \ 84 PASTEMAC(ch,axpys)( *alpha, x[i], z[i] ); \ 85 } \ 86 } \ 87 \ 88 if ( bli_is_conj( conjy ) ) \ 89 PASTEMAC(ch,conjs)( dotxy ); \ 90 \ 91 PASTEMAC(ch,copys)( dotxy, *rho ); \ 92 } \ 93 else /* bli_is_conj( conjx ) ) */ \ 94 { \ 95 conj_t conjxt_use = conjxt; \ 96 ctype dotxy; \ 97 \ 98 PASTEMAC(ch,set0s)( dotxy ); \ 99 \ 100 if ( bli_is_conj( conjy ) ) \ 101 bli_toggle_conj( &conjxt_use ); \ 102 \ 103 if ( bli_is_noconj( conjxt_use ) ) \ 104 { \ 105 PRAGMA_SIMD \ 106 for ( dim_t i = 0; i < m; ++i ) \ 107 { \ 108 PASTEMAC(ch,dots)( x[i], y[i], dotxy ); \ 109 PASTEMAC(ch,axpyjs)( *alpha, x[i], z[i] ); \ 110 } \ 111 } \ 112 else /* bli_is_conj( conjxt_use ) ) */ \ 113 { \ 114 PRAGMA_SIMD \ 115 for ( dim_t i = 0; i < m; ++i ) \ 116 { \ 117 PASTEMAC(ch,dotjs)( x[i], y[i], dotxy ); \ 118 PASTEMAC(ch,axpyjs)( *alpha, x[i], z[i] ); \ 119 } \ 120 } \ 121 \ 122 if ( bli_is_conj( conjy ) ) \ 123 PASTEMAC(ch,conjs)( dotxy ); \ 124 \ 125 PASTEMAC(ch,copys)( dotxy, *rho ); \ 126 } \ 127 } \ 128 else \ 129 { \ 130 \ 131 /* Query the context for the kernel function pointer. */ \ 132 const num_t dt = PASTEMAC(ch,type); \ 133 PASTECH(ch,dotv_ker_ft) kfp_dv \ 134 = \ 135 bli_cntx_get_l1v_ker_dt( dt, BLIS_DOTV_KER, cntx ); \ 136 PASTECH(ch,axpyv_ker_ft) kfp_av \ 137 = \ 138 bli_cntx_get_l1v_ker_dt( dt, BLIS_AXPYV_KER, cntx ); \ 139 \ 140 kfp_dv \ 141 ( \ 142 conjxt, \ 143 conjy, \ 144 m, \ 145 x, incx, \ 146 y, incy, \ 147 rho, \ 148 cntx \ 149 ); \ 150 \ 151 kfp_av \ 152 ( \ 153 conjx, \ 154 m, \ 155 alpha, \ 156 x, incx, \ 157 z, incz, \ 158 cntx \ 159 ); \ 160 } \ 161 } 162 163 INSERT_GENTFUNC_BASIC2( dotaxpyv, BLIS_CNAME_INFIX, BLIS_REF_SUFFIX ) 164 165