219 lines
5.5 KiB
C
219 lines
5.5 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 10:47:14 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 4 -dif -name hc2cbdft2_4 -include rdft/scalar/hc2cb.h */
|
|
|
|
/*
|
|
* This function contains 30 FP additions, 12 FP multiplications,
|
|
* (or, 24 additions, 6 multiplications, 6 fused multiply/add),
|
|
* 23 stack variables, 0 constants, and 16 memory accesses
|
|
*/
|
|
#include "rdft/scalar/hc2cb.h"
|
|
|
|
static void hc2cbdft2_4(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
|
|
{
|
|
{
|
|
INT m;
|
|
for (m = mb, W = W + ((mb - 1) * 6); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 6, MAKE_VOLATILE_STRIDE(16, rs)) {
|
|
E T3, Tm, T6, Tn, Td, Tk, TB, Ty, Tv, Ts;
|
|
{
|
|
E Tg, Tc, T9, Tj;
|
|
{
|
|
E T1, T2, Ta, Tb;
|
|
T1 = Rp[0];
|
|
T2 = Rm[WS(rs, 1)];
|
|
T3 = T1 + T2;
|
|
Tg = T1 - T2;
|
|
Ta = Ip[0];
|
|
Tb = Im[WS(rs, 1)];
|
|
Tc = Ta + Tb;
|
|
Tm = Ta - Tb;
|
|
}
|
|
{
|
|
E T4, T5, Th, Ti;
|
|
T4 = Rp[WS(rs, 1)];
|
|
T5 = Rm[0];
|
|
T6 = T4 + T5;
|
|
T9 = T4 - T5;
|
|
Th = Ip[WS(rs, 1)];
|
|
Ti = Im[0];
|
|
Tj = Th + Ti;
|
|
Tn = Th - Ti;
|
|
}
|
|
Td = T9 + Tc;
|
|
Tk = Tg - Tj;
|
|
TB = Tg + Tj;
|
|
Ty = Tc - T9;
|
|
Tv = Tm - Tn;
|
|
Ts = T3 - T6;
|
|
}
|
|
{
|
|
E T7, To, Te, Tp, T8, Tl, Tq, Tf;
|
|
T7 = T3 + T6;
|
|
To = Tm + Tn;
|
|
T8 = W[0];
|
|
Te = T8 * Td;
|
|
Tp = T8 * Tk;
|
|
Tf = W[1];
|
|
Tl = FMA(Tf, Tk, Te);
|
|
Tq = FNMS(Tf, Td, Tp);
|
|
Rp[0] = T7 - Tl;
|
|
Ip[0] = To + Tq;
|
|
Rm[0] = T7 + Tl;
|
|
Im[0] = Tq - To;
|
|
}
|
|
{
|
|
E Tr, Tt, Tu, TD, Tz, TF, Tx;
|
|
Tr = W[2];
|
|
Tt = Tr * Ts;
|
|
Tu = W[3];
|
|
TD = Tu * Ts;
|
|
Tx = W[4];
|
|
Tz = Tx * Ty;
|
|
TF = Tx * TB;
|
|
{
|
|
E Tw, TE, TC, TG, TA;
|
|
Tw = FNMS(Tu, Tv, Tt);
|
|
TE = FMA(Tr, Tv, TD);
|
|
TA = W[5];
|
|
TC = FMA(TA, TB, Tz);
|
|
TG = FNMS(TA, Ty, TF);
|
|
Rp[WS(rs, 1)] = Tw - TC;
|
|
Ip[WS(rs, 1)] = TE + TG;
|
|
Rm[WS(rs, 1)] = Tw + TC;
|
|
Im[WS(rs, 1)] = TG - TE;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
static const tw_instr twinstr[] = {
|
|
{ TW_FULL, 1, 4 },
|
|
{ TW_NEXT, 1, 0 }
|
|
};
|
|
|
|
static const hc2c_desc desc = { 4, "hc2cbdft2_4", twinstr, &GENUS, { 24, 6, 6, 0 } };
|
|
|
|
void X(codelet_hc2cbdft2_4) (planner *p) {
|
|
X(khc2c_register) (p, hc2cbdft2_4, &desc, HC2C_VIA_DFT);
|
|
}
|
|
#else
|
|
|
|
/* Generated by: ../../../genfft/gen_hc2cdft.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 4 -dif -name hc2cbdft2_4 -include rdft/scalar/hc2cb.h */
|
|
|
|
/*
|
|
* This function contains 30 FP additions, 12 FP multiplications,
|
|
* (or, 24 additions, 6 multiplications, 6 fused multiply/add),
|
|
* 19 stack variables, 0 constants, and 16 memory accesses
|
|
*/
|
|
#include "rdft/scalar/hc2cb.h"
|
|
|
|
static void hc2cbdft2_4(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
|
|
{
|
|
{
|
|
INT m;
|
|
for (m = mb, W = W + ((mb - 1) * 6); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 6, MAKE_VOLATILE_STRIDE(16, rs)) {
|
|
E T3, Tl, T6, Tm, Td, Tj, Tx, Tv, Ts, Tq;
|
|
{
|
|
E Tf, Tc, T9, Ti;
|
|
{
|
|
E T1, T2, Ta, Tb;
|
|
T1 = Rp[0];
|
|
T2 = Rm[WS(rs, 1)];
|
|
T3 = T1 + T2;
|
|
Tf = T1 - T2;
|
|
Ta = Ip[0];
|
|
Tb = Im[WS(rs, 1)];
|
|
Tc = Ta + Tb;
|
|
Tl = Ta - Tb;
|
|
}
|
|
{
|
|
E T4, T5, Tg, Th;
|
|
T4 = Rp[WS(rs, 1)];
|
|
T5 = Rm[0];
|
|
T6 = T4 + T5;
|
|
T9 = T4 - T5;
|
|
Tg = Ip[WS(rs, 1)];
|
|
Th = Im[0];
|
|
Ti = Tg + Th;
|
|
Tm = Tg - Th;
|
|
}
|
|
Td = T9 + Tc;
|
|
Tj = Tf - Ti;
|
|
Tx = Tf + Ti;
|
|
Tv = Tc - T9;
|
|
Ts = Tl - Tm;
|
|
Tq = T3 - T6;
|
|
}
|
|
{
|
|
E T7, Tn, Tk, To, T8, Te;
|
|
T7 = T3 + T6;
|
|
Tn = Tl + Tm;
|
|
T8 = W[0];
|
|
Te = W[1];
|
|
Tk = FMA(T8, Td, Te * Tj);
|
|
To = FNMS(Te, Td, T8 * Tj);
|
|
Rp[0] = T7 - Tk;
|
|
Ip[0] = Tn + To;
|
|
Rm[0] = T7 + Tk;
|
|
Im[0] = To - Tn;
|
|
}
|
|
{
|
|
E Tt, Tz, Ty, TA;
|
|
{
|
|
E Tp, Tr, Tu, Tw;
|
|
Tp = W[2];
|
|
Tr = W[3];
|
|
Tt = FNMS(Tr, Ts, Tp * Tq);
|
|
Tz = FMA(Tr, Tq, Tp * Ts);
|
|
Tu = W[4];
|
|
Tw = W[5];
|
|
Ty = FMA(Tu, Tv, Tw * Tx);
|
|
TA = FNMS(Tw, Tv, Tu * Tx);
|
|
}
|
|
Rp[WS(rs, 1)] = Tt - Ty;
|
|
Ip[WS(rs, 1)] = Tz + TA;
|
|
Rm[WS(rs, 1)] = Tt + Ty;
|
|
Im[WS(rs, 1)] = TA - Tz;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
static const tw_instr twinstr[] = {
|
|
{ TW_FULL, 1, 4 },
|
|
{ TW_NEXT, 1, 0 }
|
|
};
|
|
|
|
static const hc2c_desc desc = { 4, "hc2cbdft2_4", twinstr, &GENUS, { 24, 6, 6, 0 } };
|
|
|
|
void X(codelet_hc2cbdft2_4) (planner *p) {
|
|
X(khc2c_register) (p, hc2cbdft2_4, &desc, HC2C_VIA_DFT);
|
|
}
|
|
#endif
|