327 lines
8.3 KiB
C
327 lines
8.3 KiB
C
|
/*
|
||
|
|
* 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
|