Mercurial > hg > batch-feature-extraction-tool

comparison Lib/fftw-3.2.1/rdft/simd/codelets/.svn/text-base/hc2cbdftv_6.c.svn-base @ 15:585caf503ef5 tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Tidy up for ROLI
author Geogaddi\David <d.m.ronan@qmul.ac.uk>
date Tue, 17 May 2016 18:50:19 +0100
parents 636c989477e7
children
comparison
equal deleted inserted replaced
14:636c989477e7 15:585caf503ef5
1 /*
2 * Copyright (c) 2003, 2007-8 Matteo Frigo
3 * Copyright (c) 2003, 2007-8 Massachusetts Institute of Technology
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18 *
19 */
20
21 /* This file was automatically generated --- DO NOT EDIT */
22 /* Generated on Mon Feb 9 19:57:00 EST 2009 */
23
24 #include "codelet-rdft.h"
25
26 #ifdef HAVE_FMA
27
28 /* Generated by: ../../../genfft/gen_hc2cdft_c -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -trivial-stores -variables 32 -no-generate-bytw -n 6 -dif -sign 1 -name hc2cbdftv_6 -include hc2cbv.h */
29
30 /*
31 * This function contains 29 FP additions, 24 FP multiplications,
32 * (or, 17 additions, 12 multiplications, 12 fused multiply/add),
33 * 38 stack variables, 2 constants, and 12 memory accesses
34 */
35 #include "hc2cbv.h"
36
37 static void hc2cbdftv_6(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
38 {
39 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
40 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
41 INT m;
42 for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 10)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 10), MAKE_VOLATILE_STRIDE(rs)) {
43 V Tv, Tn, Tr, Te, T4, Tg, Ta, Tf, T7, T1, Td, T2, T3, T8, T9;
44 V T5, T6, Th, Tj, Tb, Tp, Tx, Ti, Tc, To, Tk, Ts, Tq, Tw, Tm;
45 V Tl, Tu, Tt, Tz, Ty;
46 T2 = LD(&(Rp[0]), ms, &(Rp[0]));
47 T3 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
48 T8 = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
49 T9 = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
50 T5 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
51 T6 = LD(&(Rm[0]), -ms, &(Rm[0]));
52 Tv = LDW(&(W[0]));
53 Tn = LDW(&(W[TWVL * 8]));
54 Tr = LDW(&(W[TWVL * 6]));
55 Te = VFMACONJ(T3, T2);
56 T4 = VFNMSCONJ(T3, T2);
57 Tg = VFMACONJ(T9, T8);
58 Ta = VFMSCONJ(T9, T8);
59 Tf = VFMACONJ(T6, T5);
60 T7 = VFNMSCONJ(T6, T5);
61 T1 = LDW(&(W[TWVL * 4]));
62 Td = LDW(&(W[TWVL * 2]));
63 Th = VADD(Tf, Tg);
64 Tj = VMUL(LDK(KP866025403), VSUB(Tf, Tg));
65 Tb = VADD(T7, Ta);
66 Tp = VMUL(LDK(KP866025403), VSUB(T7, Ta));
67 Tx = VADD(Te, Th);
68 Ti = VFNMS(LDK(KP500000000), Th, Te);
69 Tc = VZMULI(T1, VADD(T4, Tb));
70 To = VFNMS(LDK(KP500000000), Tb, T4);
71 Tk = VZMUL(Td, VFNMSI(Tj, Ti));
72 Ts = VZMUL(Tr, VFMAI(Tj, Ti));
73 Tq = VZMULI(Tn, VFNMSI(Tp, To));
74 Tw = VZMULI(Tv, VFMAI(Tp, To));
75 Tm = VCONJ(VSUB(Tk, Tc));
76 Tl = VADD(Tc, Tk);
77 Tu = VCONJ(VSUB(Ts, Tq));
78 Tt = VADD(Tq, Ts);
79 Tz = VCONJ(VSUB(Tx, Tw));
80 Ty = VADD(Tw, Tx);
81 ST(&(Rm[WS(rs, 1)]), Tm, -ms, &(Rm[WS(rs, 1)]));
82 ST(&(Rp[WS(rs, 1)]), Tl, ms, &(Rp[WS(rs, 1)]));
83 ST(&(Rm[WS(rs, 2)]), Tu, -ms, &(Rm[0]));
84 ST(&(Rp[WS(rs, 2)]), Tt, ms, &(Rp[0]));
85 ST(&(Rm[0]), Tz, -ms, &(Rm[0]));
86 ST(&(Rp[0]), Ty, ms, &(Rp[0]));
87 }
88 }
89
90 static const tw_instr twinstr[] = {
91 VTW(1, 1),
92 VTW(1, 2),
93 VTW(1, 3),
94 VTW(1, 4),
95 VTW(1, 5),
96 {TW_NEXT, VL, 0}
97 };
98
99 static const hc2c_desc desc = { 6, "hc2cbdftv_6", twinstr, &GENUS, {17, 12, 12, 0} };
100
101 void X(codelet_hc2cbdftv_6) (planner *p) {
102 X(khc2c_register) (p, hc2cbdftv_6, &desc, HC2C_VIA_DFT);
103 }
104 #else /* HAVE_FMA */
105
106 /* Generated by: ../../../genfft/gen_hc2cdft_c -simd -compact -variables 4 -pipeline-latency 8 -trivial-stores -variables 32 -no-generate-bytw -n 6 -dif -sign 1 -name hc2cbdftv_6 -include hc2cbv.h */
107
108 /*
109 * This function contains 29 FP additions, 14 FP multiplications,
110 * (or, 27 additions, 12 multiplications, 2 fused multiply/add),
111 * 41 stack variables, 2 constants, and 12 memory accesses
112 */
113 #include "hc2cbv.h"
114
115 static void hc2cbdftv_6(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
116 {
117 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
118 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
119 INT m;
120 for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 10)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 10), MAKE_VOLATILE_STRIDE(rs)) {
121 V T5, Th, Te, Ts, Tk, Tm, T2, T4, T3, T6, Tc, T8, Tb, T7, Ta;
122 V T9, Td, Ti, Tj, TA, Tf, Tn, Tv, Tt, Tz, T1, Tl, Tg, Tu, Tr;
123 V Tq, Ty, To, Tp, TC, TB, Tx, Tw;
124 T2 = LD(&(Rp[0]), ms, &(Rp[0]));
125 T3 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
126 T4 = VCONJ(T3);
127 T5 = VSUB(T2, T4);
128 Th = VADD(T2, T4);
129 T6 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
130 Tc = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
131 T7 = LD(&(Rm[0]), -ms, &(Rm[0]));
132 T8 = VCONJ(T7);
133 Ta = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
134 Tb = VCONJ(Ta);
135 T9 = VSUB(T6, T8);
136 Td = VSUB(Tb, Tc);
137 Te = VADD(T9, Td);
138 Ts = VBYI(VMUL(LDK(KP866025403), VSUB(T9, Td)));
139 Ti = VADD(T6, T8);
140 Tj = VADD(Tb, Tc);
141 Tk = VADD(Ti, Tj);
142 Tm = VBYI(VMUL(LDK(KP866025403), VSUB(Ti, Tj)));
143 TA = VADD(Th, Tk);
144 T1 = LDW(&(W[TWVL * 4]));
145 Tf = VZMULI(T1, VADD(T5, Te));
146 Tl = VFNMS(LDK(KP500000000), Tk, Th);
147 Tg = LDW(&(W[TWVL * 2]));
148 Tn = VZMUL(Tg, VSUB(Tl, Tm));
149 Tu = LDW(&(W[TWVL * 6]));
150 Tv = VZMUL(Tu, VADD(Tm, Tl));
151 Tr = VFNMS(LDK(KP500000000), Te, T5);
152 Tq = LDW(&(W[TWVL * 8]));
153 Tt = VZMULI(Tq, VSUB(Tr, Ts));
154 Ty = LDW(&(W[0]));
155 Tz = VZMULI(Ty, VADD(Ts, Tr));
156 To = VADD(Tf, Tn);
157 ST(&(Rp[WS(rs, 1)]), To, ms, &(Rp[WS(rs, 1)]));
158 Tp = VCONJ(VSUB(Tn, Tf));
159 ST(&(Rm[WS(rs, 1)]), Tp, -ms, &(Rm[WS(rs, 1)]));
160 TC = VCONJ(VSUB(TA, Tz));
161 ST(&(Rm[0]), TC, -ms, &(Rm[0]));
162 TB = VADD(Tz, TA);
163 ST(&(Rp[0]), TB, ms, &(Rp[0]));
164 Tx = VCONJ(VSUB(Tv, Tt));
165 ST(&(Rm[WS(rs, 2)]), Tx, -ms, &(Rm[0]));
166 Tw = VADD(Tt, Tv);
167 ST(&(Rp[WS(rs, 2)]), Tw, ms, &(Rp[0]));
168 }
169 }
170
171 static const tw_instr twinstr[] = {
172 VTW(1, 1),
173 VTW(1, 2),
174 VTW(1, 3),
175 VTW(1, 4),
176 VTW(1, 5),
177 {TW_NEXT, VL, 0}
178 };
179
180 static const hc2c_desc desc = { 6, "hc2cbdftv_6", twinstr, &GENUS, {27, 12, 2, 0} };
181
182 void X(codelet_hc2cbdftv_6) (planner *p) {
183 X(khc2c_register) (p, hc2cbdftv_6, &desc, HC2C_VIA_DFT);
184 }
185 #endif /* HAVE_FMA */