296 lines
7.6 KiB
C
296 lines
7.6 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 10:46:12 EDT 2021 */
|
|
|
|
#include "rdft/codelet-rdft.h"
|
|
|
|
#if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
|
|
|
|
/* Generated by: ../../../genfft/gen_hc2hc.native -fma -compact -variables 4 -pipeline-latency 4 -n 6 -dit -name hf_6 -include rdft/scalar/hf.h */
|
|
|
|
/*
|
|
* This function contains 46 FP additions, 32 FP multiplications,
|
|
* (or, 24 additions, 10 multiplications, 22 fused multiply/add),
|
|
* 31 stack variables, 2 constants, and 24 memory accesses
|
|
*/
|
|
#include "rdft/scalar/hf.h"
|
|
|
|
static void hf_6(R *cr, R *ci, 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, cr = cr + ms, ci = ci - ms, W = W + 10, MAKE_VOLATILE_STRIDE(12, rs)) {
|
|
E T1, TV, T7, TX, Tl, TR, TB, TO, Ty, TS, TC, TJ;
|
|
T1 = cr[0];
|
|
TV = ci[0];
|
|
{
|
|
E T3, T6, T4, TW, T2, T5;
|
|
T3 = cr[WS(rs, 3)];
|
|
T6 = ci[WS(rs, 3)];
|
|
T2 = W[4];
|
|
T4 = T2 * T3;
|
|
TW = T2 * T6;
|
|
T5 = W[5];
|
|
T7 = FMA(T5, T6, T4);
|
|
TX = FNMS(T5, T3, TW);
|
|
}
|
|
{
|
|
E Ta, Td, Tb, TM, Tg, Tj, Th, TK, T9, Tf;
|
|
Ta = cr[WS(rs, 2)];
|
|
Td = ci[WS(rs, 2)];
|
|
T9 = W[2];
|
|
Tb = T9 * Ta;
|
|
TM = T9 * Td;
|
|
Tg = cr[WS(rs, 5)];
|
|
Tj = ci[WS(rs, 5)];
|
|
Tf = W[8];
|
|
Th = Tf * Tg;
|
|
TK = Tf * Tj;
|
|
{
|
|
E Te, TN, Tk, TL, Tc, Ti;
|
|
Tc = W[3];
|
|
Te = FMA(Tc, Td, Tb);
|
|
TN = FNMS(Tc, Ta, TM);
|
|
Ti = W[9];
|
|
Tk = FMA(Ti, Tj, Th);
|
|
TL = FNMS(Ti, Tg, TK);
|
|
Tl = Te - Tk;
|
|
TR = TN + TL;
|
|
TB = Te + Tk;
|
|
TO = TL - TN;
|
|
}
|
|
}
|
|
{
|
|
E Tn, Tq, To, TH, Tt, Tw, Tu, TF, Tm, Ts;
|
|
Tn = cr[WS(rs, 4)];
|
|
Tq = ci[WS(rs, 4)];
|
|
Tm = W[6];
|
|
To = Tm * Tn;
|
|
TH = Tm * Tq;
|
|
Tt = cr[WS(rs, 1)];
|
|
Tw = ci[WS(rs, 1)];
|
|
Ts = W[0];
|
|
Tu = Ts * Tt;
|
|
TF = Ts * Tw;
|
|
{
|
|
E Tr, TI, Tx, TG, Tp, Tv;
|
|
Tp = W[7];
|
|
Tr = FMA(Tp, Tq, To);
|
|
TI = FNMS(Tp, Tn, TH);
|
|
Tv = W[1];
|
|
Tx = FMA(Tv, Tw, Tu);
|
|
TG = FNMS(Tv, Tt, TF);
|
|
Ty = Tr - Tx;
|
|
TS = TI + TG;
|
|
TC = Tr + Tx;
|
|
TJ = TG - TI;
|
|
}
|
|
}
|
|
{
|
|
E TP, T8, Tz, TE;
|
|
TP = TJ - TO;
|
|
T8 = T1 - T7;
|
|
Tz = Tl + Ty;
|
|
TE = FNMS(KP500000000, Tz, T8);
|
|
ci[WS(rs, 2)] = T8 + Tz;
|
|
cr[WS(rs, 1)] = FMA(KP866025403, TP, TE);
|
|
ci[0] = FNMS(KP866025403, TP, TE);
|
|
}
|
|
{
|
|
E TT, TA, TD, TQ;
|
|
TT = TR - TS;
|
|
TA = T1 + T7;
|
|
TD = TB + TC;
|
|
TQ = FNMS(KP500000000, TD, TA);
|
|
cr[0] = TA + TD;
|
|
ci[WS(rs, 1)] = FMA(KP866025403, TT, TQ);
|
|
cr[WS(rs, 2)] = FNMS(KP866025403, TT, TQ);
|
|
}
|
|
{
|
|
E T10, TU, TY, TZ;
|
|
T10 = Ty - Tl;
|
|
TU = TO + TJ;
|
|
TY = TV - TX;
|
|
TZ = FMA(KP500000000, TU, TY);
|
|
cr[WS(rs, 3)] = TU - TY;
|
|
ci[WS(rs, 4)] = FMA(KP866025403, T10, TZ);
|
|
cr[WS(rs, 5)] = FMS(KP866025403, T10, TZ);
|
|
}
|
|
{
|
|
E T14, T11, T12, T13;
|
|
T14 = TB - TC;
|
|
T11 = TX + TV;
|
|
T12 = TR + TS;
|
|
T13 = FNMS(KP500000000, T12, T11);
|
|
cr[WS(rs, 4)] = FMS(KP866025403, T14, T13);
|
|
ci[WS(rs, 5)] = T12 + T11;
|
|
ci[WS(rs, 3)] = FMA(KP866025403, T14, T13);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
static const tw_instr twinstr[] = {
|
|
{ TW_FULL, 1, 6 },
|
|
{ TW_NEXT, 1, 0 }
|
|
};
|
|
|
|
static const hc2hc_desc desc = { 6, "hf_6", twinstr, &GENUS, { 24, 10, 22, 0 } };
|
|
|
|
void X(codelet_hf_6) (planner *p) {
|
|
X(khc2hc_register) (p, hf_6, &desc);
|
|
}
|
|
#else
|
|
|
|
/* Generated by: ../../../genfft/gen_hc2hc.native -compact -variables 4 -pipeline-latency 4 -n 6 -dit -name hf_6 -include rdft/scalar/hf.h */
|
|
|
|
/*
|
|
* This function contains 46 FP additions, 28 FP multiplications,
|
|
* (or, 32 additions, 14 multiplications, 14 fused multiply/add),
|
|
* 23 stack variables, 2 constants, and 24 memory accesses
|
|
*/
|
|
#include "rdft/scalar/hf.h"
|
|
|
|
static void hf_6(R *cr, R *ci, 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, cr = cr + ms, ci = ci - ms, W = W + 10, MAKE_VOLATILE_STRIDE(12, rs)) {
|
|
E T7, TS, Tv, TO, Tt, TJ, Tx, TF, Ti, TI, Tw, TC;
|
|
{
|
|
E T1, TM, T6, TN;
|
|
T1 = cr[0];
|
|
TM = ci[0];
|
|
{
|
|
E T3, T5, T2, T4;
|
|
T3 = cr[WS(rs, 3)];
|
|
T5 = ci[WS(rs, 3)];
|
|
T2 = W[4];
|
|
T4 = W[5];
|
|
T6 = FMA(T2, T3, T4 * T5);
|
|
TN = FNMS(T4, T3, T2 * T5);
|
|
}
|
|
T7 = T1 - T6;
|
|
TS = TN + TM;
|
|
Tv = T1 + T6;
|
|
TO = TM - TN;
|
|
}
|
|
{
|
|
E Tn, TE, Ts, TD;
|
|
{
|
|
E Tk, Tm, Tj, Tl;
|
|
Tk = cr[WS(rs, 4)];
|
|
Tm = ci[WS(rs, 4)];
|
|
Tj = W[6];
|
|
Tl = W[7];
|
|
Tn = FMA(Tj, Tk, Tl * Tm);
|
|
TE = FNMS(Tl, Tk, Tj * Tm);
|
|
}
|
|
{
|
|
E Tp, Tr, To, Tq;
|
|
Tp = cr[WS(rs, 1)];
|
|
Tr = ci[WS(rs, 1)];
|
|
To = W[0];
|
|
Tq = W[1];
|
|
Ts = FMA(To, Tp, Tq * Tr);
|
|
TD = FNMS(Tq, Tp, To * Tr);
|
|
}
|
|
Tt = Tn - Ts;
|
|
TJ = TE + TD;
|
|
Tx = Tn + Ts;
|
|
TF = TD - TE;
|
|
}
|
|
{
|
|
E Tc, TA, Th, TB;
|
|
{
|
|
E T9, Tb, T8, Ta;
|
|
T9 = cr[WS(rs, 2)];
|
|
Tb = ci[WS(rs, 2)];
|
|
T8 = W[2];
|
|
Ta = W[3];
|
|
Tc = FMA(T8, T9, Ta * Tb);
|
|
TA = FNMS(Ta, T9, T8 * Tb);
|
|
}
|
|
{
|
|
E Te, Tg, Td, Tf;
|
|
Te = cr[WS(rs, 5)];
|
|
Tg = ci[WS(rs, 5)];
|
|
Td = W[8];
|
|
Tf = W[9];
|
|
Th = FMA(Td, Te, Tf * Tg);
|
|
TB = FNMS(Tf, Te, Td * Tg);
|
|
}
|
|
Ti = Tc - Th;
|
|
TI = TA + TB;
|
|
Tw = Tc + Th;
|
|
TC = TA - TB;
|
|
}
|
|
{
|
|
E TG, Tu, Tz, TK, Ty, TH;
|
|
TG = KP866025403 * (TC + TF);
|
|
Tu = Ti + Tt;
|
|
Tz = FNMS(KP500000000, Tu, T7);
|
|
ci[WS(rs, 2)] = T7 + Tu;
|
|
cr[WS(rs, 1)] = Tz + TG;
|
|
ci[0] = Tz - TG;
|
|
TK = KP866025403 * (TI - TJ);
|
|
Ty = Tw + Tx;
|
|
TH = FNMS(KP500000000, Ty, Tv);
|
|
cr[0] = Tv + Ty;
|
|
ci[WS(rs, 1)] = TH + TK;
|
|
cr[WS(rs, 2)] = TH - TK;
|
|
}
|
|
{
|
|
E TP, TL, TQ, TR, TT, TU;
|
|
TP = KP866025403 * (Tt - Ti);
|
|
TL = TF - TC;
|
|
TQ = FMA(KP500000000, TL, TO);
|
|
cr[WS(rs, 3)] = TL - TO;
|
|
ci[WS(rs, 4)] = TP + TQ;
|
|
cr[WS(rs, 5)] = TP - TQ;
|
|
TR = KP866025403 * (Tw - Tx);
|
|
TT = TI + TJ;
|
|
TU = FNMS(KP500000000, TT, TS);
|
|
cr[WS(rs, 4)] = TR - TU;
|
|
ci[WS(rs, 5)] = TT + TS;
|
|
ci[WS(rs, 3)] = TR + TU;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
static const tw_instr twinstr[] = {
|
|
{ TW_FULL, 1, 6 },
|
|
{ TW_NEXT, 1, 0 }
|
|
};
|
|
|
|
static const hc2hc_desc desc = { 6, "hf_6", twinstr, &GENUS, { 32, 14, 14, 0 } };
|
|
|
|
void X(codelet_hf_6) (planner *p) {
|
|
X(khc2hc_register) (p, hf_6, &desc);
|
|
}
|
|
#endif
|