etc
/
furnace
mirror of https://github.com/tildearrow/furnace.git
0
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
furnace/extern/fftw/dft/scalar/codelets/n1_8.c

267 lines
6.8 KiB
C
Raw Blame History

/*
* 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 10:44:24 EDT 2021 */
#include "dft/codelet-dft.h"
#if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
/* Generated by: ../../../genfft/gen_notw.native -fma -compact -variables 4 -pipeline-latency 4 -n 8 -name n1_8 -include dft/scalar/n.h */
/*
* This function contains 52 FP additions, 8 FP multiplications,
* (or, 44 additions, 0 multiplications, 8 fused multiply/add),
* 28 stack variables, 1 constants, and 32 memory accesses
*/
#include "dft/scalar/n.h"
static void n1_8(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
DK(KP707106781, +0.707106781186547524400844362104849039284835938);
{
INT i;
for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(32, is), MAKE_VOLATILE_STRIDE(32, os)) {
E T3, Tn, Ti, TC, T6, TB, Tl, To, Td, TN, Tz, TH, Ta, TM, Tu;
E TG;
{
E T1, T2, Tj, Tk;
T1 = ri[0];
T2 = ri[WS(is, 4)];
T3 = T1 + T2;
Tn = T1 - T2;
{
E Tg, Th, T4, T5;
Tg = ii[0];
Th = ii[WS(is, 4)];
Ti = Tg + Th;
TC = Tg - Th;
T4 = ri[WS(is, 2)];
T5 = ri[WS(is, 6)];
T6 = T4 + T5;
TB = T4 - T5;
}
Tj = ii[WS(is, 2)];
Tk = ii[WS(is, 6)];
Tl = Tj + Tk;
To = Tj - Tk;
{
E Tb, Tc, Tv, Tw, Tx, Ty;
Tb = ri[WS(is, 7)];
Tc = ri[WS(is, 3)];
Tv = Tb - Tc;
Tw = ii[WS(is, 7)];
Tx = ii[WS(is, 3)];
Ty = Tw - Tx;
Td = Tb + Tc;
TN = Tw + Tx;
Tz = Tv - Ty;
TH = Tv + Ty;
}
{
E T8, T9, Tq, Tr, Ts, Tt;
T8 = ri[WS(is, 1)];
T9 = ri[WS(is, 5)];
Tq = T8 - T9;
Tr = ii[WS(is, 1)];
Ts = ii[WS(is, 5)];
Tt = Tr - Ts;
Ta = T8 + T9;
TM = Tr + Ts;
Tu = Tq + Tt;
TG = Tt - Tq;
}
}
{
E T7, Te, TP, TQ;
T7 = T3 + T6;
Te = Ta + Td;
ro[WS(os, 4)] = T7 - Te;
ro[0] = T7 + Te;
TP = Ti + Tl;
TQ = TM + TN;
io[WS(os, 4)] = TP - TQ;
io[0] = TP + TQ;
}
{
E Tf, Tm, TL, TO;
Tf = Td - Ta;
Tm = Ti - Tl;
io[WS(os, 2)] = Tf + Tm;
io[WS(os, 6)] = Tm - Tf;
TL = T3 - T6;
TO = TM - TN;
ro[WS(os, 6)] = TL - TO;
ro[WS(os, 2)] = TL + TO;
}
{
E Tp, TA, TJ, TK;
Tp = Tn + To;
TA = Tu + Tz;
ro[WS(os, 5)] = FNMS(KP707106781, TA, Tp);
ro[WS(os, 1)] = FMA(KP707106781, TA, Tp);
TJ = TC - TB;
TK = TG + TH;
io[WS(os, 5)] = FNMS(KP707106781, TK, TJ);
io[WS(os, 1)] = FMA(KP707106781, TK, TJ);
}
{
E TD, TE, TF, TI;
TD = TB + TC;
TE = Tz - Tu;
io[WS(os, 7)] = FNMS(KP707106781, TE, TD);
io[WS(os, 3)] = FMA(KP707106781, TE, TD);
TF = Tn - To;
TI = TG - TH;
ro[WS(os, 7)] = FNMS(KP707106781, TI, TF);
ro[WS(os, 3)] = FMA(KP707106781, TI, TF);
}
}
}
}
static const kdft_desc desc = { 8, "n1_8", { 44, 0, 8, 0 }, &GENUS, 0, 0, 0, 0 };
void X(codelet_n1_8) (planner *p) { X(kdft_register) (p, n1_8, &desc);
}
#else
/* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 8 -name n1_8 -include dft/scalar/n.h */
/*
* This function contains 52 FP additions, 4 FP multiplications,
* (or, 52 additions, 4 multiplications, 0 fused multiply/add),
* 28 stack variables, 1 constants, and 32 memory accesses
*/
#include "dft/scalar/n.h"
static void n1_8(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
DK(KP707106781, +0.707106781186547524400844362104849039284835938);
{
INT i;
for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(32, is), MAKE_VOLATILE_STRIDE(32, os)) {
E T3, Tn, Ti, TC, T6, TB, Tl, To, Td, TN, Tz, TH, Ta, TM, Tu;
E TG;
{
E T1, T2, Tj, Tk;
T1 = ri[0];
T2 = ri[WS(is, 4)];
T3 = T1 + T2;
Tn = T1 - T2;
{
E Tg, Th, T4, T5;
Tg = ii[0];
Th = ii[WS(is, 4)];
Ti = Tg + Th;
TC = Tg - Th;
T4 = ri[WS(is, 2)];
T5 = ri[WS(is, 6)];
T6 = T4 + T5;
TB = T4 - T5;
}
Tj = ii[WS(is, 2)];
Tk = ii[WS(is, 6)];
Tl = Tj + Tk;
To = Tj - Tk;
{
E Tb, Tc, Tv, Tw, Tx, Ty;
Tb = ri[WS(is, 7)];
Tc = ri[WS(is, 3)];
Tv = Tb - Tc;
Tw = ii[WS(is, 7)];
Tx = ii[WS(is, 3)];
Ty = Tw - Tx;
Td = Tb + Tc;
TN = Tw + Tx;
Tz = Tv - Ty;
TH = Tv + Ty;
}
{
E T8, T9, Tq, Tr, Ts, Tt;
T8 = ri[WS(is, 1)];
T9 = ri[WS(is, 5)];
Tq = T8 - T9;
Tr = ii[WS(is, 1)];
Ts = ii[WS(is, 5)];
Tt = Tr - Ts;
Ta = T8 + T9;
TM = Tr + Ts;
Tu = Tq + Tt;
TG = Tt - Tq;
}
}
{
E T7, Te, TP, TQ;
T7 = T3 + T6;
Te = Ta + Td;
ro[WS(os, 4)] = T7 - Te;
ro[0] = T7 + Te;
TP = Ti + Tl;
TQ = TM + TN;
io[WS(os, 4)] = TP - TQ;
io[0] = TP + TQ;
}
{
E Tf, Tm, TL, TO;
Tf = Td - Ta;
Tm = Ti - Tl;
io[WS(os, 2)] = Tf + Tm;
io[WS(os, 6)] = Tm - Tf;
TL = T3 - T6;
TO = TM - TN;
ro[WS(os, 6)] = TL - TO;
ro[WS(os, 2)] = TL + TO;
}
{
E Tp, TA, TJ, TK;
Tp = Tn + To;
TA = KP707106781 * (Tu + Tz);
ro[WS(os, 5)] = Tp - TA;
ro[WS(os, 1)] = Tp + TA;
TJ = TC - TB;
TK = KP707106781 * (TG + TH);
io[WS(os, 5)] = TJ - TK;
io[WS(os, 1)] = TJ + TK;
}
{
E TD, TE, TF, TI;
TD = TB + TC;
TE = KP707106781 * (Tz - Tu);
io[WS(os, 7)] = TD - TE;
io[WS(os, 3)] = TD + TE;
TF = Tn - To;
TI = KP707106781 * (TG - TH);
ro[WS(os, 7)] = TF - TI;
ro[WS(os, 3)] = TF + TI;
}
}
}
}
static const kdft_desc desc = { 8, "n1_8", { 52, 4, 0, 0 }, &GENUS, 0, 0, 0, 0 };
void X(codelet_n1_8) (planner *p) { X(kdft_register) (p, n1_8, &desc);
}
#endif
Reference in New Issue View Git Blame Copy Permalink