1 /*
2 * Copyright (c) 2003, 2007-14 Matteo Frigo
3 * Copyright (c) 2003, 2007-14 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., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
18 *
19 */
20
21 /* This file was automatically generated --- DO NOT EDIT */
22 /* Generated on Thu Dec 10 07:06:25 EST 2020 */
23
24 #include "rdft/codelet-rdft.h"
25
26 #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
27
28 /* Generated by: ../../../genfft/gen_r2cb.native -fma -compact -variables 4 -pipeline-latency 4 -sign 1 -n 14 -name r2cb_14 -include rdft/scalar/r2cb.h */
29
30 /*
31 * This function contains 62 FP additions, 44 FP multiplications,
32 * (or, 18 additions, 0 multiplications, 44 fused multiply/add),
33 * 46 stack variables, 7 constants, and 28 memory accesses
34 */
35 #include "rdft/scalar/r2cb.h"
36
r2cb_14(R * R0,R * R1,R * Cr,R * Ci,stride rs,stride csr,stride csi,INT v,INT ivs,INT ovs)37 static void r2cb_14(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
38 {
39 DK(KP1_949855824, +1.949855824363647214036263365987862434465571601);
40 DK(KP1_801937735, +1.801937735804838252472204639014890102331838324);
41 DK(KP692021471, +0.692021471630095869627814897002069140197260599);
42 DK(KP801937735, +0.801937735804838252472204639014890102331838324);
43 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
44 DK(KP356895867, +0.356895867892209443894399510021300583399127187);
45 DK(KP554958132, +0.554958132087371191422194871006410481067288862);
46 {
47 INT i;
48 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(56, rs), MAKE_VOLATILE_STRIDE(56, csr), MAKE_VOLATILE_STRIDE(56, csi)) {
49 E T3, Te, To, TK, Tu, TM, Tr, TL, Tv, TA, TX, TS, TN, TF, T6;
50 E Tf, Tc, Th, T9, Tg, Tj, Tx, TU, TP, TH, TC, T1, T2, Td, Ti;
51 T1 = Cr[0];
52 T2 = Cr[WS(csr, 7)];
53 T3 = T1 - T2;
54 Te = T1 + T2;
55 {
56 E Tm, Tn, T4, T5;
57 Tm = Ci[WS(csi, 4)];
58 Tn = Ci[WS(csi, 3)];
59 To = Tm - Tn;
60 TK = Tm + Tn;
61 {
62 E Ts, Tt, Tp, Tq;
63 Ts = Ci[WS(csi, 6)];
64 Tt = Ci[WS(csi, 1)];
65 Tu = Ts - Tt;
66 TM = Ts + Tt;
67 Tp = Ci[WS(csi, 2)];
68 Tq = Ci[WS(csi, 5)];
69 Tr = Tp - Tq;
70 TL = Tp + Tq;
71 }
72 Tv = FMA(KP554958132, Tu, Tr);
73 TA = FMA(KP554958132, To, Tu);
74 TX = FNMS(KP554958132, TL, TK);
75 TS = FMA(KP554958132, TK, TM);
76 TN = FMA(KP554958132, TM, TL);
77 TF = FNMS(KP554958132, Tr, To);
78 T4 = Cr[WS(csr, 2)];
79 T5 = Cr[WS(csr, 5)];
80 T6 = T4 - T5;
81 Tf = T4 + T5;
82 {
83 E Ta, Tb, T7, T8;
84 Ta = Cr[WS(csr, 6)];
85 Tb = Cr[WS(csr, 1)];
86 Tc = Ta - Tb;
87 Th = Ta + Tb;
88 T7 = Cr[WS(csr, 4)];
89 T8 = Cr[WS(csr, 3)];
90 T9 = T7 - T8;
91 Tg = T7 + T8;
92 }
93 Tj = FNMS(KP356895867, Tg, Tf);
94 Tx = FNMS(KP356895867, Tf, Th);
95 TU = FNMS(KP356895867, Tc, T9);
96 TP = FNMS(KP356895867, T6, Tc);
97 TH = FNMS(KP356895867, T9, T6);
98 TC = FNMS(KP356895867, Th, Tg);
99 }
100 Td = T6 + T9 + Tc;
101 R1[WS(rs, 3)] = FMA(KP2_000000000, Td, T3);
102 Ti = Tf + Tg + Th;
103 R0[0] = FMA(KP2_000000000, Ti, Te);
104 {
105 E Tw, Tl, Tk, TY, TW, TV;
106 Tw = FMA(KP801937735, Tv, To);
107 Tk = FNMS(KP692021471, Tj, Th);
108 Tl = FNMS(KP1_801937735, Tk, Te);
109 R0[WS(rs, 4)] = FNMS(KP1_949855824, Tw, Tl);
110 R0[WS(rs, 3)] = FMA(KP1_949855824, Tw, Tl);
111 TY = FNMS(KP801937735, TX, TM);
112 TV = FNMS(KP692021471, TU, T6);
113 TW = FNMS(KP1_801937735, TV, T3);
114 R1[WS(rs, 1)] = FNMS(KP1_949855824, TY, TW);
115 R1[WS(rs, 5)] = FMA(KP1_949855824, TY, TW);
116 }
117 {
118 E TB, Tz, Ty, TO, TJ, TI;
119 TB = FNMS(KP801937735, TA, Tr);
120 Ty = FNMS(KP692021471, Tx, Tg);
121 Tz = FNMS(KP1_801937735, Ty, Te);
122 R0[WS(rs, 1)] = FNMS(KP1_949855824, TB, Tz);
123 R0[WS(rs, 6)] = FMA(KP1_949855824, TB, Tz);
124 TO = FMA(KP801937735, TN, TK);
125 TI = FNMS(KP692021471, TH, Tc);
126 TJ = FNMS(KP1_801937735, TI, T3);
127 R1[0] = FNMS(KP1_949855824, TO, TJ);
128 R1[WS(rs, 6)] = FMA(KP1_949855824, TO, TJ);
129 }
130 {
131 E TT, TR, TQ, TG, TE, TD;
132 TT = FNMS(KP801937735, TS, TL);
133 TQ = FNMS(KP692021471, TP, T9);
134 TR = FNMS(KP1_801937735, TQ, T3);
135 R1[WS(rs, 4)] = FNMS(KP1_949855824, TT, TR);
136 R1[WS(rs, 2)] = FMA(KP1_949855824, TT, TR);
137 TG = FNMS(KP801937735, TF, Tu);
138 TD = FNMS(KP692021471, TC, Tf);
139 TE = FNMS(KP1_801937735, TD, Te);
140 R0[WS(rs, 5)] = FNMS(KP1_949855824, TG, TE);
141 R0[WS(rs, 2)] = FMA(KP1_949855824, TG, TE);
142 }
143 }
144 }
145 }
146
147 static const kr2c_desc desc = { 14, "r2cb_14", { 18, 0, 44, 0 }, &GENUS };
148
X(codelet_r2cb_14)149 void X(codelet_r2cb_14) (planner *p) { X(kr2c_register) (p, r2cb_14, &desc);
150 }
151
152 #else
153
154 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 14 -name r2cb_14 -include rdft/scalar/r2cb.h */
155
156 /*
157 * This function contains 62 FP additions, 38 FP multiplications,
158 * (or, 36 additions, 12 multiplications, 26 fused multiply/add),
159 * 28 stack variables, 7 constants, and 28 memory accesses
160 */
161 #include "rdft/scalar/r2cb.h"
162
r2cb_14(R * R0,R * R1,R * Cr,R * Ci,stride rs,stride csr,stride csi,INT v,INT ivs,INT ovs)163 static void r2cb_14(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
164 {
165 DK(KP1_801937735, +1.801937735804838252472204639014890102331838324);
166 DK(KP445041867, +0.445041867912628808577805128993589518932711138);
167 DK(KP1_246979603, +1.246979603717467061050009768008479621264549462);
168 DK(KP867767478, +0.867767478235116240951536665696717509219981456);
169 DK(KP1_949855824, +1.949855824363647214036263365987862434465571601);
170 DK(KP1_563662964, +1.563662964936059617416889053348115500464669037);
171 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
172 {
173 INT i;
174 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(56, rs), MAKE_VOLATILE_STRIDE(56, csr), MAKE_VOLATILE_STRIDE(56, csi)) {
175 E T3, Td, T6, Te, Tq, Tz, Tn, Ty, Tc, Tg, Tk, Tx, T9, Tf, T1;
176 E T2;
177 T1 = Cr[0];
178 T2 = Cr[WS(csr, 7)];
179 T3 = T1 - T2;
180 Td = T1 + T2;
181 {
182 E T4, T5, To, Tp;
183 T4 = Cr[WS(csr, 2)];
184 T5 = Cr[WS(csr, 5)];
185 T6 = T4 - T5;
186 Te = T4 + T5;
187 To = Ci[WS(csi, 2)];
188 Tp = Ci[WS(csi, 5)];
189 Tq = To - Tp;
190 Tz = To + Tp;
191 }
192 {
193 E Tl, Tm, Ta, Tb;
194 Tl = Ci[WS(csi, 6)];
195 Tm = Ci[WS(csi, 1)];
196 Tn = Tl - Tm;
197 Ty = Tl + Tm;
198 Ta = Cr[WS(csr, 6)];
199 Tb = Cr[WS(csr, 1)];
200 Tc = Ta - Tb;
201 Tg = Ta + Tb;
202 }
203 {
204 E Ti, Tj, T7, T8;
205 Ti = Ci[WS(csi, 4)];
206 Tj = Ci[WS(csi, 3)];
207 Tk = Ti - Tj;
208 Tx = Ti + Tj;
209 T7 = Cr[WS(csr, 4)];
210 T8 = Cr[WS(csr, 3)];
211 T9 = T7 - T8;
212 Tf = T7 + T8;
213 }
214 R1[WS(rs, 3)] = FMA(KP2_000000000, T6 + T9 + Tc, T3);
215 R0[0] = FMA(KP2_000000000, Te + Tf + Tg, Td);
216 {
217 E Tr, Th, TE, TD;
218 Tr = FNMS(KP1_949855824, Tn, KP1_563662964 * Tk) - (KP867767478 * Tq);
219 Th = FMA(KP1_246979603, Tf, Td) + FNMA(KP445041867, Tg, KP1_801937735 * Te);
220 R0[WS(rs, 2)] = Th - Tr;
221 R0[WS(rs, 5)] = Th + Tr;
222 TE = FMA(KP867767478, Tx, KP1_563662964 * Ty) - (KP1_949855824 * Tz);
223 TD = FMA(KP1_246979603, Tc, T3) + FNMA(KP1_801937735, T9, KP445041867 * T6);
224 R1[WS(rs, 2)] = TD - TE;
225 R1[WS(rs, 4)] = TD + TE;
226 }
227 {
228 E Tt, Ts, TA, Tw;
229 Tt = FMA(KP867767478, Tk, KP1_563662964 * Tn) - (KP1_949855824 * Tq);
230 Ts = FMA(KP1_246979603, Tg, Td) + FNMA(KP1_801937735, Tf, KP445041867 * Te);
231 R0[WS(rs, 6)] = Ts - Tt;
232 R0[WS(rs, 1)] = Ts + Tt;
233 TA = FNMS(KP1_949855824, Ty, KP1_563662964 * Tx) - (KP867767478 * Tz);
234 Tw = FMA(KP1_246979603, T9, T3) + FNMA(KP445041867, Tc, KP1_801937735 * T6);
235 R1[WS(rs, 5)] = Tw - TA;
236 R1[WS(rs, 1)] = Tw + TA;
237 }
238 {
239 E TC, TB, Tv, Tu;
240 TC = FMA(KP1_563662964, Tz, KP1_949855824 * Tx) + (KP867767478 * Ty);
241 TB = FMA(KP1_246979603, T6, T3) + FNMA(KP1_801937735, Tc, KP445041867 * T9);
242 R1[0] = TB - TC;
243 R1[WS(rs, 6)] = TB + TC;
244 Tv = FMA(KP1_563662964, Tq, KP1_949855824 * Tk) + (KP867767478 * Tn);
245 Tu = FMA(KP1_246979603, Te, Td) + FNMA(KP1_801937735, Tg, KP445041867 * Tf);
246 R0[WS(rs, 4)] = Tu - Tv;
247 R0[WS(rs, 3)] = Tu + Tv;
248 }
249 }
250 }
251 }
252
253 static const kr2c_desc desc = { 14, "r2cb_14", { 36, 12, 26, 0 }, &GENUS };
254
X(codelet_r2cb_14)255 void X(codelet_r2cb_14) (planner *p) { X(kr2c_register) (p, r2cb_14, &desc);
256 }
257
258 #endif
259