Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/simd/common/q1bv_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 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 16:45:28 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_twidsq_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 2 -dif -name q1bv_2 -include q1b.h -sign 1 */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 6 FP additions, 4 FP multiplications,
Chris@42 32 * (or, 6 additions, 4 multiplications, 0 fused multiply/add),
Chris@42 33 * 8 stack variables, 0 constants, and 8 memory accesses
Chris@42 34 */
Chris@42 35 #include "q1b.h"
Chris@42 36
Chris@42 37 static void q1bv_2(R *ri, R *ii, const R *W, stride rs, stride vs, INT mb, INT me, INT ms)
Chris@42 38 {
Chris@42 39 {
Chris@42 40 INT m;
Chris@42 41 R *x;
Chris@42 42 x = ii;
Chris@42 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@42 44 V T1, T2, T4, T5, T3, T6;
Chris@42 45 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@42 46 T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@42 47 T4 = LD(&(x[WS(vs, 1)]), ms, &(x[WS(vs, 1)]));
Chris@42 48 T5 = LD(&(x[WS(vs, 1) + WS(rs, 1)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 49 ST(&(x[0]), VADD(T1, T2), ms, &(x[0]));
Chris@42 50 T3 = BYTW(&(W[0]), VSUB(T1, T2));
Chris@42 51 ST(&(x[WS(rs, 1)]), VADD(T4, T5), ms, &(x[WS(rs, 1)]));
Chris@42 52 T6 = BYTW(&(W[0]), VSUB(T4, T5));
Chris@42 53 ST(&(x[WS(vs, 1)]), T3, ms, &(x[WS(vs, 1)]));
Chris@42 54 ST(&(x[WS(vs, 1) + WS(rs, 1)]), T6, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 55 }
Chris@42 56 }
Chris@42 57 VLEAVE();
Chris@42 58 }
Chris@42 59
Chris@42 60 static const tw_instr twinstr[] = {
Chris@42 61 VTW(0, 1),
Chris@42 62 {TW_NEXT, VL, 0}
Chris@42 63 };
Chris@42 64
Chris@42 65 static const ct_desc desc = { 2, XSIMD_STRING("q1bv_2"), twinstr, &GENUS, {6, 4, 0, 0}, 0, 0, 0 };
Chris@42 66
Chris@42 67 void XSIMD(codelet_q1bv_2) (planner *p) {
Chris@42 68 X(kdft_difsq_register) (p, q1bv_2, &desc);
Chris@42 69 }
Chris@42 70 #else /* HAVE_FMA */
Chris@42 71
Chris@42 72 /* Generated by: ../../../genfft/gen_twidsq_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 2 -dif -name q1bv_2 -include q1b.h -sign 1 */
Chris@42 73
Chris@42 74 /*
Chris@42 75 * This function contains 6 FP additions, 4 FP multiplications,
Chris@42 76 * (or, 6 additions, 4 multiplications, 0 fused multiply/add),
Chris@42 77 * 8 stack variables, 0 constants, and 8 memory accesses
Chris@42 78 */
Chris@42 79 #include "q1b.h"
Chris@42 80
Chris@42 81 static void q1bv_2(R *ri, R *ii, const R *W, stride rs, stride vs, INT mb, INT me, INT ms)
Chris@42 82 {
Chris@42 83 {
Chris@42 84 INT m;
Chris@42 85 R *x;
Chris@42 86 x = ii;
Chris@42 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@42 88 V T1, T2, T3, T4, T5, T6;
Chris@42 89 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@42 90 T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@42 91 T3 = BYTW(&(W[0]), VSUB(T1, T2));
Chris@42 92 T4 = LD(&(x[WS(vs, 1)]), ms, &(x[WS(vs, 1)]));
Chris@42 93 T5 = LD(&(x[WS(vs, 1) + WS(rs, 1)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 94 T6 = BYTW(&(W[0]), VSUB(T4, T5));
Chris@42 95 ST(&(x[WS(vs, 1)]), T3, ms, &(x[WS(vs, 1)]));
Chris@42 96 ST(&(x[WS(vs, 1) + WS(rs, 1)]), T6, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 97 ST(&(x[0]), VADD(T1, T2), ms, &(x[0]));
Chris@42 98 ST(&(x[WS(rs, 1)]), VADD(T4, T5), ms, &(x[WS(rs, 1)]));
Chris@42 99 }
Chris@42 100 }
Chris@42 101 VLEAVE();
Chris@42 102 }
Chris@42 103
Chris@42 104 static const tw_instr twinstr[] = {
Chris@42 105 VTW(0, 1),
Chris@42 106 {TW_NEXT, VL, 0}
Chris@42 107 };
Chris@42 108
Chris@42 109 static const ct_desc desc = { 2, XSIMD_STRING("q1bv_2"), twinstr, &GENUS, {6, 4, 0, 0}, 0, 0, 0 };
Chris@42 110
Chris@42 111 void XSIMD(codelet_q1bv_2) (planner *p) {
Chris@42 112 X(kdft_difsq_register) (p, q1bv_2, &desc);
Chris@42 113 }
Chris@42 114 #endif /* HAVE_FMA */