156 lines
5.0 KiB
C
156 lines
5.0 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 10:46:47 EDT 2021 */
|
|
|
|
#include "rdft/codelet-rdft.h"
|
|
|
|
#if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
|
|
|
|
/* Generated by: ../../../genfft/gen_r2cb.native -fma -compact -variables 4 -pipeline-latency 4 -sign 1 -n 8 -name r2cb_8 -include rdft/scalar/r2cb.h */
|
|
|
|
/*
|
|
* This function contains 20 FP additions, 12 FP multiplications,
|
|
* (or, 8 additions, 0 multiplications, 12 fused multiply/add),
|
|
* 19 stack variables, 2 constants, and 16 memory accesses
|
|
*/
|
|
#include "rdft/scalar/r2cb.h"
|
|
|
|
static void r2cb_8(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
|
|
{
|
|
DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
|
|
DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
|
|
{
|
|
INT i;
|
|
for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(32, rs), MAKE_VOLATILE_STRIDE(32, csr), MAKE_VOLATILE_STRIDE(32, csi)) {
|
|
E T4, Ta, T3, T9, T8, Tc, Tf, Tk, T1, T2, T5, Tj;
|
|
T4 = Cr[WS(csr, 2)];
|
|
Ta = Ci[WS(csi, 2)];
|
|
T1 = Cr[0];
|
|
T2 = Cr[WS(csr, 4)];
|
|
T3 = T1 + T2;
|
|
T9 = T1 - T2;
|
|
{
|
|
E T6, T7, Td, Te;
|
|
T6 = Cr[WS(csr, 1)];
|
|
T7 = Cr[WS(csr, 3)];
|
|
T8 = T6 + T7;
|
|
Tc = T6 - T7;
|
|
Td = Ci[WS(csi, 1)];
|
|
Te = Ci[WS(csi, 3)];
|
|
Tf = Td + Te;
|
|
Tk = Td - Te;
|
|
}
|
|
T5 = FMA(KP2_000000000, T4, T3);
|
|
R0[WS(rs, 2)] = FNMS(KP2_000000000, T8, T5);
|
|
R0[0] = FMA(KP2_000000000, T8, T5);
|
|
Tj = FNMS(KP2_000000000, T4, T3);
|
|
R0[WS(rs, 1)] = FNMS(KP2_000000000, Tk, Tj);
|
|
R0[WS(rs, 3)] = FMA(KP2_000000000, Tk, Tj);
|
|
{
|
|
E Tb, Tg, Th, Ti;
|
|
Tb = FNMS(KP2_000000000, Ta, T9);
|
|
Tg = Tc - Tf;
|
|
R1[WS(rs, 2)] = FNMS(KP1_414213562, Tg, Tb);
|
|
R1[0] = FMA(KP1_414213562, Tg, Tb);
|
|
Th = FMA(KP2_000000000, Ta, T9);
|
|
Ti = Tc + Tf;
|
|
R1[WS(rs, 1)] = FNMS(KP1_414213562, Ti, Th);
|
|
R1[WS(rs, 3)] = FMA(KP1_414213562, Ti, Th);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
static const kr2c_desc desc = { 8, "r2cb_8", { 8, 0, 12, 0 }, &GENUS };
|
|
|
|
void X(codelet_r2cb_8) (planner *p) { X(kr2c_register) (p, r2cb_8, &desc);
|
|
}
|
|
|
|
#else
|
|
|
|
/* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 8 -name r2cb_8 -include rdft/scalar/r2cb.h */
|
|
|
|
/*
|
|
* This function contains 20 FP additions, 6 FP multiplications,
|
|
* (or, 20 additions, 6 multiplications, 0 fused multiply/add),
|
|
* 21 stack variables, 2 constants, and 16 memory accesses
|
|
*/
|
|
#include "rdft/scalar/r2cb.h"
|
|
|
|
static void r2cb_8(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
|
|
{
|
|
DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
|
|
DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
|
|
{
|
|
INT i;
|
|
for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(32, rs), MAKE_VOLATILE_STRIDE(32, csr), MAKE_VOLATILE_STRIDE(32, csi)) {
|
|
E T5, Tg, T3, Te, T9, Ti, Td, Tj, T6, Ta;
|
|
{
|
|
E T4, Tf, T1, T2;
|
|
T4 = Cr[WS(csr, 2)];
|
|
T5 = KP2_000000000 * T4;
|
|
Tf = Ci[WS(csi, 2)];
|
|
Tg = KP2_000000000 * Tf;
|
|
T1 = Cr[0];
|
|
T2 = Cr[WS(csr, 4)];
|
|
T3 = T1 + T2;
|
|
Te = T1 - T2;
|
|
{
|
|
E T7, T8, Tb, Tc;
|
|
T7 = Cr[WS(csr, 1)];
|
|
T8 = Cr[WS(csr, 3)];
|
|
T9 = KP2_000000000 * (T7 + T8);
|
|
Ti = T7 - T8;
|
|
Tb = Ci[WS(csi, 1)];
|
|
Tc = Ci[WS(csi, 3)];
|
|
Td = KP2_000000000 * (Tb - Tc);
|
|
Tj = Tb + Tc;
|
|
}
|
|
}
|
|
T6 = T3 + T5;
|
|
R0[WS(rs, 2)] = T6 - T9;
|
|
R0[0] = T6 + T9;
|
|
Ta = T3 - T5;
|
|
R0[WS(rs, 1)] = Ta - Td;
|
|
R0[WS(rs, 3)] = Ta + Td;
|
|
{
|
|
E Th, Tk, Tl, Tm;
|
|
Th = Te - Tg;
|
|
Tk = KP1_414213562 * (Ti - Tj);
|
|
R1[WS(rs, 2)] = Th - Tk;
|
|
R1[0] = Th + Tk;
|
|
Tl = Te + Tg;
|
|
Tm = KP1_414213562 * (Ti + Tj);
|
|
R1[WS(rs, 1)] = Tl - Tm;
|
|
R1[WS(rs, 3)] = Tl + Tm;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
static const kr2c_desc desc = { 8, "r2cb_8", { 20, 6, 0, 0 }, &GENUS };
|
|
|
|
void X(codelet_r2cb_8) (planner *p) { X(kr2c_register) (p, r2cb_8, &desc);
|
|
}
|
|
|
|
#endif
|