mirror of
https://github.com/tildearrow/furnace.git
synced 2024-11-06 12:55:05 +00:00
54e93db207
not reliable yet
426 lines
13 KiB
C
426 lines
13 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 10:45:49 EDT 2021 */
|
|
|
|
#include "dft/codelet-dft.h"
|
|
|
|
#if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
|
|
|
|
/* Generated by: ../../../genfft/gen_twiddle_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -n 16 -name t1bv_16 -include dft/simd/t1b.h -sign 1 */
|
|
|
|
/*
|
|
* This function contains 87 FP additions, 64 FP multiplications,
|
|
* (or, 53 additions, 30 multiplications, 34 fused multiply/add),
|
|
* 36 stack variables, 3 constants, and 32 memory accesses
|
|
*/
|
|
#include "dft/simd/t1b.h"
|
|
|
|
static void t1bv_16(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
|
|
{
|
|
DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
|
|
DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
|
|
DVK(KP414213562, +0.414213562373095048801688724209698078569671875);
|
|
{
|
|
INT m;
|
|
R *x;
|
|
x = ii;
|
|
for (m = mb, W = W + (mb * ((TWVL / VL) * 30)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 30), MAKE_VOLATILE_STRIDE(16, rs)) {
|
|
V T4, TW, T9, T19, TD, TI, TZ, T1a, Tf, Tk, Tl, T13, T1c, Tq, Tv;
|
|
V Tw, T16, T1d, T1, T3, T2;
|
|
T1 = LD(&(x[0]), ms, &(x[0]));
|
|
T2 = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
|
|
T3 = BYTW(&(W[TWVL * 14]), T2);
|
|
T4 = VADD(T1, T3);
|
|
TW = VSUB(T1, T3);
|
|
{
|
|
V T6, T8, T5, T7;
|
|
T5 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
|
|
T6 = BYTW(&(W[TWVL * 6]), T5);
|
|
T7 = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
|
|
T8 = BYTW(&(W[TWVL * 22]), T7);
|
|
T9 = VADD(T6, T8);
|
|
T19 = VSUB(T6, T8);
|
|
}
|
|
{
|
|
V TA, TH, TC, TF, TX, TY;
|
|
{
|
|
V Tz, TG, TB, TE;
|
|
Tz = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
|
|
TA = BYTW(&(W[TWVL * 2]), Tz);
|
|
TG = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
|
|
TH = BYTW(&(W[TWVL * 10]), TG);
|
|
TB = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
|
|
TC = BYTW(&(W[TWVL * 18]), TB);
|
|
TE = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
|
|
TF = BYTW(&(W[TWVL * 26]), TE);
|
|
}
|
|
TD = VADD(TA, TC);
|
|
TI = VADD(TF, TH);
|
|
TX = VSUB(TA, TC);
|
|
TY = VSUB(TF, TH);
|
|
TZ = VADD(TX, TY);
|
|
T1a = VSUB(TX, TY);
|
|
}
|
|
{
|
|
V Tc, Tj, Te, Th, T11, T12;
|
|
{
|
|
V Tb, Ti, Td, Tg;
|
|
Tb = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
|
|
Tc = BYTW(&(W[0]), Tb);
|
|
Ti = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
|
|
Tj = BYTW(&(W[TWVL * 24]), Ti);
|
|
Td = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
|
|
Te = BYTW(&(W[TWVL * 16]), Td);
|
|
Tg = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
|
|
Th = BYTW(&(W[TWVL * 8]), Tg);
|
|
}
|
|
Tf = VADD(Tc, Te);
|
|
Tk = VADD(Th, Tj);
|
|
Tl = VSUB(Tf, Tk);
|
|
T11 = VSUB(Tc, Te);
|
|
T12 = VSUB(Th, Tj);
|
|
T13 = VFNMS(LDK(KP414213562), T12, T11);
|
|
T1c = VFMA(LDK(KP414213562), T11, T12);
|
|
}
|
|
{
|
|
V Tn, Tu, Tp, Ts, T14, T15;
|
|
{
|
|
V Tm, Tt, To, Tr;
|
|
Tm = LD(&(x[WS(rs, 15)]), ms, &(x[WS(rs, 1)]));
|
|
Tn = BYTW(&(W[TWVL * 28]), Tm);
|
|
Tt = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
|
|
Tu = BYTW(&(W[TWVL * 20]), Tt);
|
|
To = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
|
|
Tp = BYTW(&(W[TWVL * 12]), To);
|
|
Tr = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
|
|
Ts = BYTW(&(W[TWVL * 4]), Tr);
|
|
}
|
|
Tq = VADD(Tn, Tp);
|
|
Tv = VADD(Ts, Tu);
|
|
Tw = VSUB(Tq, Tv);
|
|
T14 = VSUB(Tn, Tp);
|
|
T15 = VSUB(Tu, Ts);
|
|
T16 = VFNMS(LDK(KP414213562), T15, T14);
|
|
T1d = VFMA(LDK(KP414213562), T14, T15);
|
|
}
|
|
{
|
|
V Ty, TM, TL, TN;
|
|
{
|
|
V Ta, Tx, TJ, TK;
|
|
Ta = VSUB(T4, T9);
|
|
Tx = VADD(Tl, Tw);
|
|
Ty = VFNMS(LDK(KP707106781), Tx, Ta);
|
|
TM = VFMA(LDK(KP707106781), Tx, Ta);
|
|
TJ = VSUB(TD, TI);
|
|
TK = VSUB(Tl, Tw);
|
|
TL = VFNMS(LDK(KP707106781), TK, TJ);
|
|
TN = VFMA(LDK(KP707106781), TK, TJ);
|
|
}
|
|
ST(&(x[WS(rs, 6)]), VFNMSI(TL, Ty), ms, &(x[0]));
|
|
ST(&(x[WS(rs, 14)]), VFNMSI(TN, TM), ms, &(x[0]));
|
|
ST(&(x[WS(rs, 10)]), VFMAI(TL, Ty), ms, &(x[0]));
|
|
ST(&(x[WS(rs, 2)]), VFMAI(TN, TM), ms, &(x[0]));
|
|
}
|
|
{
|
|
V T1k, T1o, T1n, T1p;
|
|
{
|
|
V T1i, T1j, T1l, T1m;
|
|
T1i = VFNMS(LDK(KP707106781), TZ, TW);
|
|
T1j = VADD(T1c, T1d);
|
|
T1k = VFNMS(LDK(KP923879532), T1j, T1i);
|
|
T1o = VFMA(LDK(KP923879532), T1j, T1i);
|
|
T1l = VFNMS(LDK(KP707106781), T1a, T19);
|
|
T1m = VSUB(T13, T16);
|
|
T1n = VFMA(LDK(KP923879532), T1m, T1l);
|
|
T1p = VFNMS(LDK(KP923879532), T1m, T1l);
|
|
}
|
|
ST(&(x[WS(rs, 5)]), VFMAI(T1n, T1k), ms, &(x[WS(rs, 1)]));
|
|
ST(&(x[WS(rs, 13)]), VFMAI(T1p, T1o), ms, &(x[WS(rs, 1)]));
|
|
ST(&(x[WS(rs, 11)]), VFNMSI(T1n, T1k), ms, &(x[WS(rs, 1)]));
|
|
ST(&(x[WS(rs, 3)]), VFNMSI(T1p, T1o), ms, &(x[WS(rs, 1)]));
|
|
}
|
|
{
|
|
V TQ, TU, TT, TV;
|
|
{
|
|
V TO, TP, TR, TS;
|
|
TO = VADD(T4, T9);
|
|
TP = VADD(TD, TI);
|
|
TQ = VSUB(TO, TP);
|
|
TU = VADD(TO, TP);
|
|
TR = VADD(Tf, Tk);
|
|
TS = VADD(Tq, Tv);
|
|
TT = VSUB(TR, TS);
|
|
TV = VADD(TR, TS);
|
|
}
|
|
ST(&(x[WS(rs, 12)]), VFNMSI(TT, TQ), ms, &(x[0]));
|
|
ST(&(x[0]), VADD(TU, TV), ms, &(x[0]));
|
|
ST(&(x[WS(rs, 4)]), VFMAI(TT, TQ), ms, &(x[0]));
|
|
ST(&(x[WS(rs, 8)]), VSUB(TU, TV), ms, &(x[0]));
|
|
}
|
|
{
|
|
V T18, T1g, T1f, T1h;
|
|
{
|
|
V T10, T17, T1b, T1e;
|
|
T10 = VFMA(LDK(KP707106781), TZ, TW);
|
|
T17 = VADD(T13, T16);
|
|
T18 = VFNMS(LDK(KP923879532), T17, T10);
|
|
T1g = VFMA(LDK(KP923879532), T17, T10);
|
|
T1b = VFMA(LDK(KP707106781), T1a, T19);
|
|
T1e = VSUB(T1c, T1d);
|
|
T1f = VFNMS(LDK(KP923879532), T1e, T1b);
|
|
T1h = VFMA(LDK(KP923879532), T1e, T1b);
|
|
}
|
|
ST(&(x[WS(rs, 7)]), VFNMSI(T1f, T18), ms, &(x[WS(rs, 1)]));
|
|
ST(&(x[WS(rs, 1)]), VFMAI(T1h, T1g), ms, &(x[WS(rs, 1)]));
|
|
ST(&(x[WS(rs, 9)]), VFMAI(T1f, T18), ms, &(x[WS(rs, 1)]));
|
|
ST(&(x[WS(rs, 15)]), VFNMSI(T1h, T1g), ms, &(x[WS(rs, 1)]));
|
|
}
|
|
}
|
|
}
|
|
VLEAVE();
|
|
}
|
|
|
|
static const tw_instr twinstr[] = {
|
|
VTW(0, 1),
|
|
VTW(0, 2),
|
|
VTW(0, 3),
|
|
VTW(0, 4),
|
|
VTW(0, 5),
|
|
VTW(0, 6),
|
|
VTW(0, 7),
|
|
VTW(0, 8),
|
|
VTW(0, 9),
|
|
VTW(0, 10),
|
|
VTW(0, 11),
|
|
VTW(0, 12),
|
|
VTW(0, 13),
|
|
VTW(0, 14),
|
|
VTW(0, 15),
|
|
{ TW_NEXT, VL, 0 }
|
|
};
|
|
|
|
static const ct_desc desc = { 16, XSIMD_STRING("t1bv_16"), twinstr, &GENUS, { 53, 30, 34, 0 }, 0, 0, 0 };
|
|
|
|
void XSIMD(codelet_t1bv_16) (planner *p) {
|
|
X(kdft_dit_register) (p, t1bv_16, &desc);
|
|
}
|
|
#else
|
|
|
|
/* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 16 -name t1bv_16 -include dft/simd/t1b.h -sign 1 */
|
|
|
|
/*
|
|
* This function contains 87 FP additions, 42 FP multiplications,
|
|
* (or, 83 additions, 38 multiplications, 4 fused multiply/add),
|
|
* 36 stack variables, 3 constants, and 32 memory accesses
|
|
*/
|
|
#include "dft/simd/t1b.h"
|
|
|
|
static void t1bv_16(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
|
|
{
|
|
DVK(KP382683432, +0.382683432365089771728459984030398866761344562);
|
|
DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
|
|
DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
|
|
{
|
|
INT m;
|
|
R *x;
|
|
x = ii;
|
|
for (m = mb, W = W + (mb * ((TWVL / VL) * 30)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 30), MAKE_VOLATILE_STRIDE(16, rs)) {
|
|
V TJ, T1b, TD, T1c, T17, T18, Ty, TK, T10, T11, T12, Tb, TM, T13, T14;
|
|
V T15, Tm, TN, TG, TI, TH;
|
|
TG = LD(&(x[0]), ms, &(x[0]));
|
|
TH = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
|
|
TI = BYTW(&(W[TWVL * 14]), TH);
|
|
TJ = VSUB(TG, TI);
|
|
T1b = VADD(TG, TI);
|
|
{
|
|
V TA, TC, Tz, TB;
|
|
Tz = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
|
|
TA = BYTW(&(W[TWVL * 6]), Tz);
|
|
TB = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
|
|
TC = BYTW(&(W[TWVL * 22]), TB);
|
|
TD = VSUB(TA, TC);
|
|
T1c = VADD(TA, TC);
|
|
}
|
|
{
|
|
V Tp, Tw, Tr, Tu, Ts, Tx;
|
|
{
|
|
V To, Tv, Tq, Tt;
|
|
To = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
|
|
Tp = BYTW(&(W[TWVL * 2]), To);
|
|
Tv = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
|
|
Tw = BYTW(&(W[TWVL * 10]), Tv);
|
|
Tq = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
|
|
Tr = BYTW(&(W[TWVL * 18]), Tq);
|
|
Tt = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
|
|
Tu = BYTW(&(W[TWVL * 26]), Tt);
|
|
}
|
|
T17 = VADD(Tp, Tr);
|
|
T18 = VADD(Tu, Tw);
|
|
Ts = VSUB(Tp, Tr);
|
|
Tx = VSUB(Tu, Tw);
|
|
Ty = VMUL(LDK(KP707106781), VSUB(Ts, Tx));
|
|
TK = VMUL(LDK(KP707106781), VADD(Ts, Tx));
|
|
}
|
|
{
|
|
V T2, T9, T4, T7, T5, Ta;
|
|
{
|
|
V T1, T8, T3, T6;
|
|
T1 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
|
|
T2 = BYTW(&(W[0]), T1);
|
|
T8 = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
|
|
T9 = BYTW(&(W[TWVL * 24]), T8);
|
|
T3 = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
|
|
T4 = BYTW(&(W[TWVL * 16]), T3);
|
|
T6 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
|
|
T7 = BYTW(&(W[TWVL * 8]), T6);
|
|
}
|
|
T10 = VADD(T2, T4);
|
|
T11 = VADD(T7, T9);
|
|
T12 = VSUB(T10, T11);
|
|
T5 = VSUB(T2, T4);
|
|
Ta = VSUB(T7, T9);
|
|
Tb = VFNMS(LDK(KP382683432), Ta, VMUL(LDK(KP923879532), T5));
|
|
TM = VFMA(LDK(KP382683432), T5, VMUL(LDK(KP923879532), Ta));
|
|
}
|
|
{
|
|
V Td, Tk, Tf, Ti, Tg, Tl;
|
|
{
|
|
V Tc, Tj, Te, Th;
|
|
Tc = LD(&(x[WS(rs, 15)]), ms, &(x[WS(rs, 1)]));
|
|
Td = BYTW(&(W[TWVL * 28]), Tc);
|
|
Tj = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
|
|
Tk = BYTW(&(W[TWVL * 20]), Tj);
|
|
Te = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
|
|
Tf = BYTW(&(W[TWVL * 12]), Te);
|
|
Th = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
|
|
Ti = BYTW(&(W[TWVL * 4]), Th);
|
|
}
|
|
T13 = VADD(Td, Tf);
|
|
T14 = VADD(Ti, Tk);
|
|
T15 = VSUB(T13, T14);
|
|
Tg = VSUB(Td, Tf);
|
|
Tl = VSUB(Ti, Tk);
|
|
Tm = VFMA(LDK(KP923879532), Tg, VMUL(LDK(KP382683432), Tl));
|
|
TN = VFNMS(LDK(KP382683432), Tg, VMUL(LDK(KP923879532), Tl));
|
|
}
|
|
{
|
|
V T1a, T1g, T1f, T1h;
|
|
{
|
|
V T16, T19, T1d, T1e;
|
|
T16 = VMUL(LDK(KP707106781), VSUB(T12, T15));
|
|
T19 = VSUB(T17, T18);
|
|
T1a = VBYI(VSUB(T16, T19));
|
|
T1g = VBYI(VADD(T19, T16));
|
|
T1d = VSUB(T1b, T1c);
|
|
T1e = VMUL(LDK(KP707106781), VADD(T12, T15));
|
|
T1f = VSUB(T1d, T1e);
|
|
T1h = VADD(T1d, T1e);
|
|
}
|
|
ST(&(x[WS(rs, 6)]), VADD(T1a, T1f), ms, &(x[0]));
|
|
ST(&(x[WS(rs, 14)]), VSUB(T1h, T1g), ms, &(x[0]));
|
|
ST(&(x[WS(rs, 10)]), VSUB(T1f, T1a), ms, &(x[0]));
|
|
ST(&(x[WS(rs, 2)]), VADD(T1g, T1h), ms, &(x[0]));
|
|
}
|
|
{
|
|
V T1k, T1o, T1n, T1p;
|
|
{
|
|
V T1i, T1j, T1l, T1m;
|
|
T1i = VADD(T1b, T1c);
|
|
T1j = VADD(T17, T18);
|
|
T1k = VSUB(T1i, T1j);
|
|
T1o = VADD(T1i, T1j);
|
|
T1l = VADD(T10, T11);
|
|
T1m = VADD(T13, T14);
|
|
T1n = VBYI(VSUB(T1l, T1m));
|
|
T1p = VADD(T1l, T1m);
|
|
}
|
|
ST(&(x[WS(rs, 12)]), VSUB(T1k, T1n), ms, &(x[0]));
|
|
ST(&(x[0]), VADD(T1o, T1p), ms, &(x[0]));
|
|
ST(&(x[WS(rs, 4)]), VADD(T1k, T1n), ms, &(x[0]));
|
|
ST(&(x[WS(rs, 8)]), VSUB(T1o, T1p), ms, &(x[0]));
|
|
}
|
|
{
|
|
V TF, TQ, TP, TR;
|
|
{
|
|
V Tn, TE, TL, TO;
|
|
Tn = VSUB(Tb, Tm);
|
|
TE = VSUB(Ty, TD);
|
|
TF = VBYI(VSUB(Tn, TE));
|
|
TQ = VBYI(VADD(TE, Tn));
|
|
TL = VSUB(TJ, TK);
|
|
TO = VSUB(TM, TN);
|
|
TP = VSUB(TL, TO);
|
|
TR = VADD(TL, TO);
|
|
}
|
|
ST(&(x[WS(rs, 5)]), VADD(TF, TP), ms, &(x[WS(rs, 1)]));
|
|
ST(&(x[WS(rs, 13)]), VSUB(TR, TQ), ms, &(x[WS(rs, 1)]));
|
|
ST(&(x[WS(rs, 11)]), VSUB(TP, TF), ms, &(x[WS(rs, 1)]));
|
|
ST(&(x[WS(rs, 3)]), VADD(TQ, TR), ms, &(x[WS(rs, 1)]));
|
|
}
|
|
{
|
|
V TU, TY, TX, TZ;
|
|
{
|
|
V TS, TT, TV, TW;
|
|
TS = VADD(TJ, TK);
|
|
TT = VADD(Tb, Tm);
|
|
TU = VADD(TS, TT);
|
|
TY = VSUB(TS, TT);
|
|
TV = VADD(TD, Ty);
|
|
TW = VADD(TM, TN);
|
|
TX = VBYI(VADD(TV, TW));
|
|
TZ = VBYI(VSUB(TW, TV));
|
|
}
|
|
ST(&(x[WS(rs, 15)]), VSUB(TU, TX), ms, &(x[WS(rs, 1)]));
|
|
ST(&(x[WS(rs, 7)]), VADD(TY, TZ), ms, &(x[WS(rs, 1)]));
|
|
ST(&(x[WS(rs, 1)]), VADD(TU, TX), ms, &(x[WS(rs, 1)]));
|
|
ST(&(x[WS(rs, 9)]), VSUB(TY, TZ), ms, &(x[WS(rs, 1)]));
|
|
}
|
|
}
|
|
}
|
|
VLEAVE();
|
|
}
|
|
|
|
static const tw_instr twinstr[] = {
|
|
VTW(0, 1),
|
|
VTW(0, 2),
|
|
VTW(0, 3),
|
|
VTW(0, 4),
|
|
VTW(0, 5),
|
|
VTW(0, 6),
|
|
VTW(0, 7),
|
|
VTW(0, 8),
|
|
VTW(0, 9),
|
|
VTW(0, 10),
|
|
VTW(0, 11),
|
|
VTW(0, 12),
|
|
VTW(0, 13),
|
|
VTW(0, 14),
|
|
VTW(0, 15),
|
|
{ TW_NEXT, VL, 0 }
|
|
};
|
|
|
|
static const ct_desc desc = { 16, XSIMD_STRING("t1bv_16"), twinstr, &GENUS, { 83, 38, 4, 0 }, 0, 0, 0 };
|
|
|
|
void XSIMD(codelet_t1bv_16) (planner *p) {
|
|
X(kdft_dit_register) (p, t1bv_16, &desc);
|
|
}
|
|
#endif
|