Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/dft/simd/common/n2bv_4.c @ 82:d0c2a83c1364

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add FFTW 3.3.8 source, and a Linux build
author Chris Cannam
date Tue, 19 Nov 2019 14:52:55 +0000
parents
children
rev   line source
Chris@82 1 /*
Chris@82 2 * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@82 3 * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@82 4 *
Chris@82 5 * This program is free software; you can redistribute it and/or modify
Chris@82 6 * it under the terms of the GNU General Public License as published by
Chris@82 7 * the Free Software Foundation; either version 2 of the License, or
Chris@82 8 * (at your option) any later version.
Chris@82 9 *
Chris@82 10 * This program is distributed in the hope that it will be useful,
Chris@82 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@82 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
Chris@82 13 * GNU General Public License for more details.
Chris@82 14 *
Chris@82 15 * You should have received a copy of the GNU General Public License
Chris@82 16 * along with this program; if not, write to the Free Software
Chris@82 17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Chris@82 18 *
Chris@82 19 */
Chris@82 20
Chris@82 21 /* This file was automatically generated --- DO NOT EDIT */
Chris@82 22 /* Generated on Thu May 24 08:05:11 EDT 2018 */
Chris@82 23
Chris@82 24 #include "dft/codelet-dft.h"
Chris@82 25
Chris@82 26 #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
Chris@82 27
Chris@82 28 /* Generated by: ../../../genfft/gen_notw_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 4 -name n2bv_4 -with-ostride 2 -include dft/simd/n2b.h -store-multiple 2 */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 8 FP additions, 2 FP multiplications,
Chris@82 32 * (or, 6 additions, 0 multiplications, 2 fused multiply/add),
Chris@82 33 * 11 stack variables, 0 constants, and 10 memory accesses
Chris@82 34 */
Chris@82 35 #include "dft/simd/n2b.h"
Chris@82 36
Chris@82 37 static void n2bv_4(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@82 38 {
Chris@82 39 {
Chris@82 40 INT i;
Chris@82 41 const R *xi;
Chris@82 42 R *xo;
Chris@82 43 xi = ii;
Chris@82 44 xo = io;
Chris@82 45 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(8, is), MAKE_VOLATILE_STRIDE(8, os)) {
Chris@82 46 V T3, T7, T6, T8;
Chris@82 47 {
Chris@82 48 V T1, T2, T4, T5;
Chris@82 49 T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@82 50 T2 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@82 51 T3 = VSUB(T1, T2);
Chris@82 52 T7 = VADD(T1, T2);
Chris@82 53 T4 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@82 54 T5 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@82 55 T6 = VSUB(T4, T5);
Chris@82 56 T8 = VADD(T4, T5);
Chris@82 57 }
Chris@82 58 {
Chris@82 59 V T9, Ta, Tb, Tc;
Chris@82 60 T9 = VFNMSI(T6, T3);
Chris@82 61 STM2(&(xo[6]), T9, ovs, &(xo[2]));
Chris@82 62 Ta = VADD(T7, T8);
Chris@82 63 STM2(&(xo[0]), Ta, ovs, &(xo[0]));
Chris@82 64 Tb = VFMAI(T6, T3);
Chris@82 65 STM2(&(xo[2]), Tb, ovs, &(xo[2]));
Chris@82 66 STN2(&(xo[0]), Ta, Tb, ovs);
Chris@82 67 Tc = VSUB(T7, T8);
Chris@82 68 STM2(&(xo[4]), Tc, ovs, &(xo[0]));
Chris@82 69 STN2(&(xo[4]), Tc, T9, ovs);
Chris@82 70 }
Chris@82 71 }
Chris@82 72 }
Chris@82 73 VLEAVE();
Chris@82 74 }
Chris@82 75
Chris@82 76 static const kdft_desc desc = { 4, XSIMD_STRING("n2bv_4"), {6, 0, 2, 0}, &GENUS, 0, 2, 0, 0 };
Chris@82 77
Chris@82 78 void XSIMD(codelet_n2bv_4) (planner *p) {
Chris@82 79 X(kdft_register) (p, n2bv_4, &desc);
Chris@82 80 }
Chris@82 81
Chris@82 82 #else
Chris@82 83
Chris@82 84 /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 4 -name n2bv_4 -with-ostride 2 -include dft/simd/n2b.h -store-multiple 2 */
Chris@82 85
Chris@82 86 /*
Chris@82 87 * This function contains 8 FP additions, 0 FP multiplications,
Chris@82 88 * (or, 8 additions, 0 multiplications, 0 fused multiply/add),
Chris@82 89 * 11 stack variables, 0 constants, and 10 memory accesses
Chris@82 90 */
Chris@82 91 #include "dft/simd/n2b.h"
Chris@82 92
Chris@82 93 static void n2bv_4(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@82 94 {
Chris@82 95 {
Chris@82 96 INT i;
Chris@82 97 const R *xi;
Chris@82 98 R *xo;
Chris@82 99 xi = ii;
Chris@82 100 xo = io;
Chris@82 101 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(8, is), MAKE_VOLATILE_STRIDE(8, os)) {
Chris@82 102 V T3, T7, T6, T8;
Chris@82 103 {
Chris@82 104 V T1, T2, T4, T5;
Chris@82 105 T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@82 106 T2 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@82 107 T3 = VSUB(T1, T2);
Chris@82 108 T7 = VADD(T1, T2);
Chris@82 109 T4 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@82 110 T5 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@82 111 T6 = VBYI(VSUB(T4, T5));
Chris@82 112 T8 = VADD(T4, T5);
Chris@82 113 }
Chris@82 114 {
Chris@82 115 V T9, Ta, Tb, Tc;
Chris@82 116 T9 = VSUB(T3, T6);
Chris@82 117 STM2(&(xo[6]), T9, ovs, &(xo[2]));
Chris@82 118 Ta = VADD(T7, T8);
Chris@82 119 STM2(&(xo[0]), Ta, ovs, &(xo[0]));
Chris@82 120 Tb = VADD(T3, T6);
Chris@82 121 STM2(&(xo[2]), Tb, ovs, &(xo[2]));
Chris@82 122 STN2(&(xo[0]), Ta, Tb, ovs);
Chris@82 123 Tc = VSUB(T7, T8);
Chris@82 124 STM2(&(xo[4]), Tc, ovs, &(xo[0]));
Chris@82 125 STN2(&(xo[4]), Tc, T9, ovs);
Chris@82 126 }
Chris@82 127 }
Chris@82 128 }
Chris@82 129 VLEAVE();
Chris@82 130 }
Chris@82 131
Chris@82 132 static const kdft_desc desc = { 4, XSIMD_STRING("n2bv_4"), {8, 0, 0, 0}, &GENUS, 0, 2, 0, 0 };
Chris@82 133
Chris@82 134 void XSIMD(codelet_n2bv_4) (planner *p) {
Chris@82 135 X(kdft_register) (p, n2bv_4, &desc);
Chris@82 136 }
Chris@82 137
Chris@82 138 #endif