alexander
/
iup-stack
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases 1 Wiki Activity
iup-stack/fftw/rdft/scalar/r2cb/hc2cbdft_6.c

327 lines
8.3 KiB
C
Raw Permalink Blame History

/*
* Copyright (c) 2003, 2007-14 Matteo Frigo
* Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*
*/
/* This file was automatically generated --- DO NOT EDIT */
/* Generated on Tue Sep 14 10:47:12 EDT 2021 */
#include "rdft/codelet-rdft.h"
#if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
/* Generated by: ../../../genfft/gen_hc2cdft.native -fma -compact -variables 4 -pipeline-latency 4 -sign 1 -n 6 -dif -name hc2cbdft_6 -include rdft/scalar/hc2cb.h */
/*
* This function contains 58 FP additions, 32 FP multiplications,
* (or, 36 additions, 10 multiplications, 22 fused multiply/add),
* 34 stack variables, 2 constants, and 24 memory accesses
*/
#include "rdft/scalar/hc2cb.h"
static void hc2cbdft_6(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
DK(KP866025403, +0.866025403784438646763723170752936183471402627);
DK(KP500000000, +0.500000000000000000000000000000000000000000000);
{
INT m;
for (m = mb, W = W + ((mb - 1) * 10); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 10, MAKE_VOLATILE_STRIDE(24, rs)) {
E Tp, TD, Tj, TV, Tq, Tr, TG, TP, T4, Ts, TQ, Tb, Tc, TA, TU;
{
E Tf, TF, Ti, TE, Td, Te;
Td = Ip[WS(rs, 1)];
Te = Im[WS(rs, 1)];
Tf = Td - Te;
TF = Te + Td;
{
E Tn, To, Tg, Th;
Tn = Ip[0];
To = Im[WS(rs, 2)];
Tp = Tn - To;
TD = Tn + To;
Tg = Ip[WS(rs, 2)];
Th = Im[0];
Ti = Tg - Th;
TE = Tg + Th;
}
Tj = Tf - Ti;
TV = TF + TE;
Tq = Tf + Ti;
Tr = FNMS(KP500000000, Tq, Tp);
TG = TE - TF;
TP = FNMS(KP500000000, TG, TD);
}
{
E Tw, Ta, Ty, T7, Tx, T2, T3, Tz;
T2 = Rp[0];
T3 = Rm[WS(rs, 2)];
T4 = T2 + T3;
Tw = T2 - T3;
{
E T8, T9, T5, T6;
T8 = Rm[WS(rs, 1)];
T9 = Rp[WS(rs, 1)];
Ta = T8 + T9;
Ty = T8 - T9;
T5 = Rp[WS(rs, 2)];
T6 = Rm[0];
T7 = T5 + T6;
Tx = T5 - T6;
}
Ts = T7 - Ta;
TQ = Tx - Ty;
Tb = T7 + Ta;
Tc = FNMS(KP500000000, Tb, T4);
Tz = Tx + Ty;
TA = Tw + Tz;
TU = FNMS(KP500000000, Tz, Tw);
}
{
E TN, TY, TR, TW, TS, TZ, TO, TX, T10, TT;
TN = T4 + Tb;
TY = Tp + Tq;
TR = FMA(KP866025403, TQ, TP);
TW = FNMS(KP866025403, TV, TU);
TO = W[0];
TS = TO * TR;
TZ = TO * TW;
TT = W[1];
TX = FMA(TT, TW, TS);
T10 = FNMS(TT, TR, TZ);
Rp[0] = TN - TX;
Ip[0] = TY + T10;
Rm[0] = TN + TX;
Im[0] = T10 - TY;
}
{
E Tt, TH, Tv, TB, TC, TL, T1, Tl, Tm, TJ, Tk;
Tt = FNMS(KP866025403, Ts, Tr);
TH = TD + TG;
Tv = W[4];
TB = Tv * TA;
TC = W[5];
TL = TC * TA;
Tk = FNMS(KP866025403, Tj, Tc);
T1 = W[3];
Tl = T1 * Tk;
Tm = W[2];
TJ = Tm * Tk;
{
E Tu, TI, TK, TM;
Tu = FMA(Tm, Tt, Tl);
TI = FNMS(TC, TH, TB);
Ip[WS(rs, 1)] = Tu + TI;
Im[WS(rs, 1)] = TI - Tu;
TK = FNMS(T1, Tt, TJ);
TM = FMA(Tv, TH, TL);
Rp[WS(rs, 1)] = TK - TM;
Rm[WS(rs, 1)] = TK + TM;
}
}
{
E T15, T11, T13, T14, T1d, T18, T1b, T19, T1f, T12, T17;
T15 = FMA(KP866025403, Ts, Tr);
T12 = FMA(KP866025403, Tj, Tc);
T11 = W[6];
T13 = T11 * T12;
T14 = W[7];
T1d = T14 * T12;
T18 = FNMS(KP866025403, TQ, TP);
T1b = FMA(KP866025403, TV, TU);
T17 = W[8];
T19 = T17 * T18;
T1f = T17 * T1b;
{
E T16, T1e, T1c, T1g, T1a;
T16 = FNMS(T14, T15, T13);
T1e = FMA(T11, T15, T1d);
T1a = W[9];
T1c = FMA(T1a, T1b, T19);
T1g = FNMS(T1a, T18, T1f);
Rp[WS(rs, 2)] = T16 - T1c;
Ip[WS(rs, 2)] = T1e + T1g;
Rm[WS(rs, 2)] = T16 + T1c;
Im[WS(rs, 2)] = T1g - T1e;
}
}
}
}
}
static const tw_instr twinstr[] = {
{ TW_FULL, 1, 6 },
{ TW_NEXT, 1, 0 }
};
static const hc2c_desc desc = { 6, "hc2cbdft_6", twinstr, &GENUS, { 36, 10, 22, 0 } };
void X(codelet_hc2cbdft_6) (planner *p) {
X(khc2c_register) (p, hc2cbdft_6, &desc, HC2C_VIA_DFT);
}
#else
/* Generated by: ../../../genfft/gen_hc2cdft.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 6 -dif -name hc2cbdft_6 -include rdft/scalar/hc2cb.h */
/*
* This function contains 58 FP additions, 28 FP multiplications,
* (or, 44 additions, 14 multiplications, 14 fused multiply/add),
* 29 stack variables, 2 constants, and 24 memory accesses
*/
#include "rdft/scalar/hc2cb.h"
static void hc2cbdft_6(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
DK(KP500000000, +0.500000000000000000000000000000000000000000000);
DK(KP866025403, +0.866025403784438646763723170752936183471402627);
{
INT m;
for (m = mb, W = W + ((mb - 1) * 10); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 10, MAKE_VOLATILE_STRIDE(24, rs)) {
E T4, Tv, Tr, TL, Tb, Tc, Ty, TP, To, TB, Tj, TQ, Tp, Tq, TE;
E TM;
{
E Ta, Tx, T7, Tw, T2, T3;
T2 = Rp[0];
T3 = Rm[WS(rs, 2)];
T4 = T2 + T3;
Tv = T2 - T3;
{
E T8, T9, T5, T6;
T8 = Rm[WS(rs, 1)];
T9 = Rp[WS(rs, 1)];
Ta = T8 + T9;
Tx = T8 - T9;
T5 = Rp[WS(rs, 2)];
T6 = Rm[0];
T7 = T5 + T6;
Tw = T5 - T6;
}
Tr = KP866025403 * (T7 - Ta);
TL = KP866025403 * (Tw - Tx);
Tb = T7 + Ta;
Tc = FNMS(KP500000000, Tb, T4);
Ty = Tw + Tx;
TP = FNMS(KP500000000, Ty, Tv);
}
{
E Tf, TC, Ti, TD, Td, Te;
Td = Ip[WS(rs, 1)];
Te = Im[WS(rs, 1)];
Tf = Td - Te;
TC = Te + Td;
{
E Tm, Tn, Tg, Th;
Tm = Ip[0];
Tn = Im[WS(rs, 2)];
To = Tm - Tn;
TB = Tm + Tn;
Tg = Ip[WS(rs, 2)];
Th = Im[0];
Ti = Tg - Th;
TD = Tg + Th;
}
Tj = KP866025403 * (Tf - Ti);
TQ = KP866025403 * (TC + TD);
Tp = Tf + Ti;
Tq = FNMS(KP500000000, Tp, To);
TE = TC - TD;
TM = FMA(KP500000000, TE, TB);
}
{
E TJ, TT, TS, TU;
TJ = T4 + Tb;
TT = To + Tp;
{
E TN, TR, TK, TO;
TN = TL + TM;
TR = TP - TQ;
TK = W[0];
TO = W[1];
TS = FMA(TK, TN, TO * TR);
TU = FNMS(TO, TN, TK * TR);
}
Rp[0] = TJ - TS;
Ip[0] = TT + TU;
Rm[0] = TJ + TS;
Im[0] = TU - TT;
}
{
E TZ, T15, T14, T16;
{
E TW, TY, TV, TX;
TW = Tc + Tj;
TY = Tr + Tq;
TV = W[6];
TX = W[7];
TZ = FNMS(TX, TY, TV * TW);
T15 = FMA(TX, TW, TV * TY);
}
{
E T11, T13, T10, T12;
T11 = TM - TL;
T13 = TP + TQ;
T10 = W[8];
T12 = W[9];
T14 = FMA(T10, T11, T12 * T13);
T16 = FNMS(T12, T11, T10 * T13);
}
Rp[WS(rs, 2)] = TZ - T14;
Ip[WS(rs, 2)] = T15 + T16;
Rm[WS(rs, 2)] = TZ + T14;
Im[WS(rs, 2)] = T16 - T15;
}
{
E Tt, TH, TG, TI;
{
E Tk, Ts, T1, Tl;
Tk = Tc - Tj;
Ts = Tq - Tr;
T1 = W[3];
Tl = W[2];
Tt = FMA(T1, Tk, Tl * Ts);
TH = FNMS(T1, Ts, Tl * Tk);
}
{
E Tz, TF, Tu, TA;
Tz = Tv + Ty;
TF = TB - TE;
Tu = W[4];
TA = W[5];
TG = FNMS(TA, TF, Tu * Tz);
TI = FMA(TA, Tz, Tu * TF);
}
Ip[WS(rs, 1)] = Tt + TG;
Rp[WS(rs, 1)] = TH - TI;
Im[WS(rs, 1)] = TG - Tt;
Rm[WS(rs, 1)] = TH + TI;
}
}
}
}
static const tw_instr twinstr[] = {
{ TW_FULL, 1, 6 },
{ TW_NEXT, 1, 0 }
};
static const hc2c_desc desc = { 6, "hc2cbdft_6", twinstr, &GENUS, { 44, 14, 14, 0 } };
void X(codelet_hc2cbdft_6) (planner *p) {
X(khc2c_register) (p, hc2cbdft_6, &desc, HC2C_VIA_DFT);
}
#endif
Reference in New Issue View Git Blame Copy Permalink