201 lines
5.2 KiB
C
201 lines
5.2 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 10:46:18 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 -twiddle-log3 -precompute-twiddles -n 4 -dit -name hf2_4 -include rdft/scalar/hf.h */
|
|
|
|
/*
|
|
* This function contains 24 FP additions, 16 FP multiplications,
|
|
* (or, 16 additions, 8 multiplications, 8 fused multiply/add),
|
|
* 21 stack variables, 0 constants, and 16 memory accesses
|
|
*/
|
|
#include "rdft/scalar/hf.h"
|
|
|
|
static void hf2_4(R *cr, R *ci, const R *W, stride rs, INT mb, INT me, INT ms)
|
|
{
|
|
{
|
|
INT m;
|
|
for (m = mb, W = W + ((mb - 1) * 4); m < me; m = m + 1, cr = cr + ms, ci = ci - ms, W = W + 4, MAKE_VOLATILE_STRIDE(8, rs)) {
|
|
E T2, T6, T3, T5, T7, Tb, T4, Ta;
|
|
T2 = W[0];
|
|
T6 = W[3];
|
|
T3 = W[2];
|
|
T4 = T2 * T3;
|
|
Ta = T2 * T6;
|
|
T5 = W[1];
|
|
T7 = FMA(T5, T6, T4);
|
|
Tb = FNMS(T5, T3, Ta);
|
|
{
|
|
E T1, Tx, Td, Tw, Ti, Tq, Tm, Ts;
|
|
T1 = cr[0];
|
|
Tx = ci[0];
|
|
{
|
|
E T8, T9, Tc, Tv;
|
|
T8 = cr[WS(rs, 2)];
|
|
T9 = T7 * T8;
|
|
Tc = ci[WS(rs, 2)];
|
|
Tv = T7 * Tc;
|
|
Td = FMA(Tb, Tc, T9);
|
|
Tw = FNMS(Tb, T8, Tv);
|
|
}
|
|
{
|
|
E Tf, Tg, Th, Tp;
|
|
Tf = cr[WS(rs, 1)];
|
|
Tg = T2 * Tf;
|
|
Th = ci[WS(rs, 1)];
|
|
Tp = T2 * Th;
|
|
Ti = FMA(T5, Th, Tg);
|
|
Tq = FNMS(T5, Tf, Tp);
|
|
}
|
|
{
|
|
E Tj, Tk, Tl, Tr;
|
|
Tj = cr[WS(rs, 3)];
|
|
Tk = T3 * Tj;
|
|
Tl = ci[WS(rs, 3)];
|
|
Tr = T3 * Tl;
|
|
Tm = FMA(T6, Tl, Tk);
|
|
Ts = FNMS(T6, Tj, Tr);
|
|
}
|
|
{
|
|
E Te, Tn, To, Tt;
|
|
Te = T1 + Td;
|
|
Tn = Ti + Tm;
|
|
ci[WS(rs, 1)] = Te - Tn;
|
|
cr[0] = Te + Tn;
|
|
To = T1 - Td;
|
|
Tt = Tq - Ts;
|
|
ci[0] = To - Tt;
|
|
cr[WS(rs, 1)] = To + Tt;
|
|
}
|
|
{
|
|
E Tu, Ty, Tz, TA;
|
|
Tu = Tq + Ts;
|
|
Ty = Tw + Tx;
|
|
cr[WS(rs, 2)] = Tu - Ty;
|
|
ci[WS(rs, 3)] = Tu + Ty;
|
|
Tz = Tm - Ti;
|
|
TA = Tx - Tw;
|
|
cr[WS(rs, 3)] = Tz - TA;
|
|
ci[WS(rs, 2)] = Tz + TA;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
static const tw_instr twinstr[] = {
|
|
{ TW_CEXP, 1, 1 },
|
|
{ TW_CEXP, 1, 3 },
|
|
{ TW_NEXT, 1, 0 }
|
|
};
|
|
|
|
static const hc2hc_desc desc = { 4, "hf2_4", twinstr, &GENUS, { 16, 8, 8, 0 } };
|
|
|
|
void X(codelet_hf2_4) (planner *p) {
|
|
X(khc2hc_register) (p, hf2_4, &desc);
|
|
}
|
|
#else
|
|
|
|
/* Generated by: ../../../genfft/gen_hc2hc.native -compact -variables 4 -pipeline-latency 4 -twiddle-log3 -precompute-twiddles -n 4 -dit -name hf2_4 -include rdft/scalar/hf.h */
|
|
|
|
/*
|
|
* This function contains 24 FP additions, 16 FP multiplications,
|
|
* (or, 16 additions, 8 multiplications, 8 fused multiply/add),
|
|
* 21 stack variables, 0 constants, and 16 memory accesses
|
|
*/
|
|
#include "rdft/scalar/hf.h"
|
|
|
|
static void hf2_4(R *cr, R *ci, const R *W, stride rs, INT mb, INT me, INT ms)
|
|
{
|
|
{
|
|
INT m;
|
|
for (m = mb, W = W + ((mb - 1) * 4); m < me; m = m + 1, cr = cr + ms, ci = ci - ms, W = W + 4, MAKE_VOLATILE_STRIDE(8, rs)) {
|
|
E T2, T4, T3, T5, T6, T8;
|
|
T2 = W[0];
|
|
T4 = W[1];
|
|
T3 = W[2];
|
|
T5 = W[3];
|
|
T6 = FMA(T2, T3, T4 * T5);
|
|
T8 = FNMS(T4, T3, T2 * T5);
|
|
{
|
|
E T1, Tp, Ta, To, Te, Tk, Th, Tl, T7, T9;
|
|
T1 = cr[0];
|
|
Tp = ci[0];
|
|
T7 = cr[WS(rs, 2)];
|
|
T9 = ci[WS(rs, 2)];
|
|
Ta = FMA(T6, T7, T8 * T9);
|
|
To = FNMS(T8, T7, T6 * T9);
|
|
{
|
|
E Tc, Td, Tf, Tg;
|
|
Tc = cr[WS(rs, 1)];
|
|
Td = ci[WS(rs, 1)];
|
|
Te = FMA(T2, Tc, T4 * Td);
|
|
Tk = FNMS(T4, Tc, T2 * Td);
|
|
Tf = cr[WS(rs, 3)];
|
|
Tg = ci[WS(rs, 3)];
|
|
Th = FMA(T3, Tf, T5 * Tg);
|
|
Tl = FNMS(T5, Tf, T3 * Tg);
|
|
}
|
|
{
|
|
E Tb, Ti, Tj, Tm;
|
|
Tb = T1 + Ta;
|
|
Ti = Te + Th;
|
|
ci[WS(rs, 1)] = Tb - Ti;
|
|
cr[0] = Tb + Ti;
|
|
Tj = T1 - Ta;
|
|
Tm = Tk - Tl;
|
|
ci[0] = Tj - Tm;
|
|
cr[WS(rs, 1)] = Tj + Tm;
|
|
}
|
|
{
|
|
E Tn, Tq, Tr, Ts;
|
|
Tn = Tk + Tl;
|
|
Tq = To + Tp;
|
|
cr[WS(rs, 2)] = Tn - Tq;
|
|
ci[WS(rs, 3)] = Tn + Tq;
|
|
Tr = Th - Te;
|
|
Ts = Tp - To;
|
|
cr[WS(rs, 3)] = Tr - Ts;
|
|
ci[WS(rs, 2)] = Tr + Ts;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
static const tw_instr twinstr[] = {
|
|
{ TW_CEXP, 1, 1 },
|
|
{ TW_CEXP, 1, 3 },
|
|
{ TW_NEXT, 1, 0 }
|
|
};
|
|
|
|
static const hc2hc_desc desc = { 4, "hf2_4", twinstr, &GENUS, { 16, 8, 8, 0 } };
|
|
|
|
void X(codelet_hf2_4) (planner *p) {
|
|
X(khc2hc_register) (p, hf2_4, &desc);
|
|
}
|
|
#endif
|