Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/dft/simd/common/q1fv_2.c @ 84:08ae793730bd

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add null config files
author Chris Cannam
date Mon, 02 Mar 2020 14:03:47 +0000
parents d0c2a83c1364
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 08:06:13 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_twidsq_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -n 2 -dif -name q1fv_2 -include dft/simd/q1f.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 6 FP additions, 4 FP multiplications,
Chris@82 32 * (or, 6 additions, 4 multiplications, 0 fused multiply/add),
Chris@82 33 * 8 stack variables, 0 constants, and 8 memory accesses
Chris@82 34 */
Chris@82 35 #include "dft/simd/q1f.h"
Chris@82 36
Chris@82 37 static void q1fv_2(R *ri, R *ii, const R *W, stride rs, stride vs, INT mb, INT me, INT ms)
Chris@82 38 {
Chris@82 39 {
Chris@82 40 INT m;
Chris@82 41 R *x;
Chris@82 42 x = ri;
Chris@82 43 for (m = mb, W = W + (mb * ((TWVL / VL) * 2)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 2), MAKE_VOLATILE_STRIDE(4, rs), MAKE_VOLATILE_STRIDE(4, vs)) {
Chris@82 44 V T1, T2, T3, T4, T5, T6;
Chris@82 45 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@82 46 T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@82 47 T3 = BYTWJ(&(W[0]), VSUB(T1, T2));
Chris@82 48 T4 = LD(&(x[WS(vs, 1)]), ms, &(x[WS(vs, 1)]));
Chris@82 49 T5 = LD(&(x[WS(vs, 1) + WS(rs, 1)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@82 50 T6 = BYTWJ(&(W[0]), VSUB(T4, T5));
Chris@82 51 ST(&(x[WS(vs, 1)]), T3, ms, &(x[WS(vs, 1)]));
Chris@82 52 ST(&(x[WS(vs, 1) + WS(rs, 1)]), T6, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@82 53 ST(&(x[0]), VADD(T1, T2), ms, &(x[0]));
Chris@82 54 ST(&(x[WS(rs, 1)]), VADD(T4, T5), ms, &(x[WS(rs, 1)]));
Chris@82 55 }
Chris@82 56 }
Chris@82 57 VLEAVE();
Chris@82 58 }
Chris@82 59
Chris@82 60 static const tw_instr twinstr[] = {
Chris@82 61 VTW(0, 1),
Chris@82 62 {TW_NEXT, VL, 0}
Chris@82 63 };
Chris@82 64
Chris@82 65 static const ct_desc desc = { 2, XSIMD_STRING("q1fv_2"), twinstr, &GENUS, {6, 4, 0, 0}, 0, 0, 0 };
Chris@82 66
Chris@82 67 void XSIMD(codelet_q1fv_2) (planner *p) {
Chris@82 68 X(kdft_difsq_register) (p, q1fv_2, &desc);
Chris@82 69 }
Chris@82 70 #else
Chris@82 71
Chris@82 72 /* Generated by: ../../../genfft/gen_twidsq_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 2 -dif -name q1fv_2 -include dft/simd/q1f.h */
Chris@82 73
Chris@82 74 /*
Chris@82 75 * This function contains 6 FP additions, 4 FP multiplications,
Chris@82 76 * (or, 6 additions, 4 multiplications, 0 fused multiply/add),
Chris@82 77 * 8 stack variables, 0 constants, and 8 memory accesses
Chris@82 78 */
Chris@82 79 #include "dft/simd/q1f.h"
Chris@82 80
Chris@82 81 static void q1fv_2(R *ri, R *ii, const R *W, stride rs, stride vs, INT mb, INT me, INT ms)
Chris@82 82 {
Chris@82 83 {
Chris@82 84 INT m;
Chris@82 85 R *x;
Chris@82 86 x = ri;
Chris@82 87 for (m = mb, W = W + (mb * ((TWVL / VL) * 2)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 2), MAKE_VOLATILE_STRIDE(4, rs), MAKE_VOLATILE_STRIDE(4, vs)) {
Chris@82 88 V T1, T2, T3, T4, T5, T6;
Chris@82 89 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@82 90 T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@82 91 T3 = BYTWJ(&(W[0]), VSUB(T1, T2));
Chris@82 92 T4 = LD(&(x[WS(vs, 1)]), ms, &(x[WS(vs, 1)]));
Chris@82 93 T5 = LD(&(x[WS(vs, 1) + WS(rs, 1)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@82 94 T6 = BYTWJ(&(W[0]), VSUB(T4, T5));
Chris@82 95 ST(&(x[WS(vs, 1)]), T3, ms, &(x[WS(vs, 1)]));
Chris@82 96 ST(&(x[WS(vs, 1) + WS(rs, 1)]), T6, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@82 97 ST(&(x[0]), VADD(T1, T2), ms, &(x[0]));
Chris@82 98 ST(&(x[WS(rs, 1)]), VADD(T4, T5), ms, &(x[WS(rs, 1)]));
Chris@82 99 }
Chris@82 100 }
Chris@82 101 VLEAVE();
Chris@82 102 }
Chris@82 103
Chris@82 104 static const tw_instr twinstr[] = {
Chris@82 105 VTW(0, 1),
Chris@82 106 {TW_NEXT, VL, 0}
Chris@82 107 };
Chris@82 108
Chris@82 109 static const ct_desc desc = { 2, XSIMD_STRING("q1fv_2"), twinstr, &GENUS, {6, 4, 0, 0}, 0, 0, 0 };
Chris@82 110
Chris@82 111 void XSIMD(codelet_q1fv_2) (planner *p) {
Chris@82 112 X(kdft_difsq_register) (p, q1fv_2, &desc);
Chris@82 113 }
Chris@82 114 #endif