annotate src/fftw-3.3.5/dft/simd/common/t1bv_3.c @ 84:08ae793730bd

Add null config files
author Chris Cannam
date Mon, 02 Mar 2020 14:03:47 +0000
parents 2cd0e3b3e1fd
children
rev   line source
Chris@42 1 /*
Chris@42 2 * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@42 3 * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@42 4 *
Chris@42 5 * This program is free software; you can redistribute it and/or modify
Chris@42 6 * it under the terms of the GNU General Public License as published by
Chris@42 7 * the Free Software Foundation; either version 2 of the License, or
Chris@42 8 * (at your option) any later version.
Chris@42 9 *
Chris@42 10 * This program is distributed in the hope that it will be useful,
Chris@42 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@42 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
Chris@42 13 * GNU General Public License for more details.
Chris@42 14 *
Chris@42 15 * You should have received a copy of the GNU General Public License
Chris@42 16 * along with this program; if not, write to the Free Software
Chris@42 17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Chris@42 18 *
Chris@42 19 */
Chris@42 20
Chris@42 21 /* This file was automatically generated --- DO NOT EDIT */
Chris@42 22 /* Generated on Sat Jul 30 16:44:14 EDT 2016 */
Chris@42 23
Chris@42 24 #include "codelet-dft.h"
Chris@42 25
Chris@42 26 #ifdef HAVE_FMA
Chris@42 27
Chris@42 28 /* Generated by: ../../../genfft/gen_twiddle_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 3 -name t1bv_3 -include t1b.h -sign 1 */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 8 FP additions, 8 FP multiplications,
Chris@42 32 * (or, 5 additions, 5 multiplications, 3 fused multiply/add),
Chris@42 33 * 12 stack variables, 2 constants, and 6 memory accesses
Chris@42 34 */
Chris@42 35 #include "t1b.h"
Chris@42 36
Chris@42 37 static void t1bv_3(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42 38 {
Chris@42 39 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42 40 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42 41 {
Chris@42 42 INT m;
Chris@42 43 R *x;
Chris@42 44 x = ii;
Chris@42 45 for (m = mb, W = W + (mb * ((TWVL / VL) * 4)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 4), MAKE_VOLATILE_STRIDE(3, rs)) {
Chris@42 46 V T1, T2, T4;
Chris@42 47 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@42 48 T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@42 49 T4 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@42 50 {
Chris@42 51 V T3, T5, T8, T6, T7;
Chris@42 52 T3 = BYTW(&(W[0]), T2);
Chris@42 53 T5 = BYTW(&(W[TWVL * 2]), T4);
Chris@42 54 T8 = VMUL(LDK(KP866025403), VSUB(T3, T5));
Chris@42 55 T6 = VADD(T3, T5);
Chris@42 56 T7 = VFNMS(LDK(KP500000000), T6, T1);
Chris@42 57 ST(&(x[0]), VADD(T1, T6), ms, &(x[0]));
Chris@42 58 ST(&(x[WS(rs, 2)]), VFNMSI(T8, T7), ms, &(x[0]));
Chris@42 59 ST(&(x[WS(rs, 1)]), VFMAI(T8, T7), ms, &(x[WS(rs, 1)]));
Chris@42 60 }
Chris@42 61 }
Chris@42 62 }
Chris@42 63 VLEAVE();
Chris@42 64 }
Chris@42 65
Chris@42 66 static const tw_instr twinstr[] = {
Chris@42 67 VTW(0, 1),
Chris@42 68 VTW(0, 2),
Chris@42 69 {TW_NEXT, VL, 0}
Chris@42 70 };
Chris@42 71
Chris@42 72 static const ct_desc desc = { 3, XSIMD_STRING("t1bv_3"), twinstr, &GENUS, {5, 5, 3, 0}, 0, 0, 0 };
Chris@42 73
Chris@42 74 void XSIMD(codelet_t1bv_3) (planner *p) {
Chris@42 75 X(kdft_dit_register) (p, t1bv_3, &desc);
Chris@42 76 }
Chris@42 77 #else /* HAVE_FMA */
Chris@42 78
Chris@42 79 /* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 3 -name t1bv_3 -include t1b.h -sign 1 */
Chris@42 80
Chris@42 81 /*
Chris@42 82 * This function contains 8 FP additions, 6 FP multiplications,
Chris@42 83 * (or, 7 additions, 5 multiplications, 1 fused multiply/add),
Chris@42 84 * 12 stack variables, 2 constants, and 6 memory accesses
Chris@42 85 */
Chris@42 86 #include "t1b.h"
Chris@42 87
Chris@42 88 static void t1bv_3(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42 89 {
Chris@42 90 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42 91 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42 92 {
Chris@42 93 INT m;
Chris@42 94 R *x;
Chris@42 95 x = ii;
Chris@42 96 for (m = mb, W = W + (mb * ((TWVL / VL) * 4)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 4), MAKE_VOLATILE_STRIDE(3, rs)) {
Chris@42 97 V T6, T2, T4, T7, T1, T3, T5, T8;
Chris@42 98 T6 = LD(&(x[0]), ms, &(x[0]));
Chris@42 99 T1 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@42 100 T2 = BYTW(&(W[0]), T1);
Chris@42 101 T3 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@42 102 T4 = BYTW(&(W[TWVL * 2]), T3);
Chris@42 103 T7 = VADD(T2, T4);
Chris@42 104 ST(&(x[0]), VADD(T6, T7), ms, &(x[0]));
Chris@42 105 T5 = VBYI(VMUL(LDK(KP866025403), VSUB(T2, T4)));
Chris@42 106 T8 = VFNMS(LDK(KP500000000), T7, T6);
Chris@42 107 ST(&(x[WS(rs, 1)]), VADD(T5, T8), ms, &(x[WS(rs, 1)]));
Chris@42 108 ST(&(x[WS(rs, 2)]), VSUB(T8, T5), ms, &(x[0]));
Chris@42 109 }
Chris@42 110 }
Chris@42 111 VLEAVE();
Chris@42 112 }
Chris@42 113
Chris@42 114 static const tw_instr twinstr[] = {
Chris@42 115 VTW(0, 1),
Chris@42 116 VTW(0, 2),
Chris@42 117 {TW_NEXT, VL, 0}
Chris@42 118 };
Chris@42 119
Chris@42 120 static const ct_desc desc = { 3, XSIMD_STRING("t1bv_3"), twinstr, &GENUS, {7, 5, 1, 0}, 0, 0, 0 };
Chris@42 121
Chris@42 122 void XSIMD(codelet_t1bv_3) (planner *p) {
Chris@42 123 X(kdft_dit_register) (p, t1bv_3, &desc);
Chris@42 124 }
Chris@42 125 #endif /* HAVE_FMA */