etc/furnace
0
0
Fork 0
mirror of https://github.com/tildearrow/furnace.git synced 2024-12-18 22:40:20 +00:00
Code Issues Projects Releases Packages Wiki Activity
furnace/extern/fftw/rdft/simd/common/hc2cbdftv_12.c
tildearrow 54e93db207 GUI: try using FFTW for per-chan osc wave center 
not reliable yet
2022-05-31 03:24:29 -05:00

327 lines
12 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 10:47:22 EDT 2021 */
#include "rdft/codelet-rdft.h"
#if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
/* Generated by: ../../../genfft/gen_hc2cdft_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -trivial-stores -variables 32 -no-generate-bytw -n 12 -dif -sign 1 -name hc2cbdftv_12 -include rdft/simd/hc2cbv.h */
/*
* This function contains 71 FP additions, 51 FP multiplications,
* (or, 45 additions, 25 multiplications, 26 fused multiply/add),
* 56 stack variables, 2 constants, and 24 memory accesses
*/
#include "rdft/simd/hc2cbv.h"
static void hc2cbdftv_12(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
{
INT m;
for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 22)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 22), MAKE_VOLATILE_STRIDE(48, rs)) {
V Tk, Tw, Td, TA, T11, T1f, TF, TP, Tt, TB, TY, T1e;
{
V T2, Tm, T7, T8, Tp, Tq, T5, Tu, Tg, Tr, Tj, Tn, Tb, Tv, T3;
V T4, Te, Tf, Th, Ti, T9, Ta, T6, Tc, TZ, T10, TD, TE, To, Ts;
V TW, TX;
T2 = LD(&(Rp[0]), ms, &(Rp[0]));
Tm = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
T7 = LD(&(Rm[WS(rs, 5)]), -ms, &(Rm[WS(rs, 1)]));
T8 = VCONJ(T7);
Tp = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
Tq = VCONJ(Tp);
T3 = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
T4 = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
T5 = VFMACONJ(T4, T3);
Tu = VFNMSCONJ(T4, T3);
Te = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
Tf = LD(&(Rp[WS(rs, 5)]), ms, &(Rp[WS(rs, 1)]));
Tg = VSUB(Te, Tf);
Tr = VADD(Te, Tf);
Th = LD(&(Rm[0]), -ms, &(Rm[0]));
Ti = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
Tj = VSUB(Th, Ti);
Tn = VADD(Ti, Th);
T9 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
Ta = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
Tb = VFMACONJ(Ta, T9);
Tv = VFMSCONJ(Ta, T9);
Tk = VFMACONJ(Tj, Tg);
Tw = VSUB(Tu, Tv);
T6 = VFNMS(LDK(KP500000000), T5, T2);
Tc = VFNMS(LDK(KP500000000), Tb, T8);
Td = VSUB(T6, Tc);
TA = VADD(T6, Tc);
TZ = VFMACONJ(Tn, Tm);
T10 = VFMACONJ(Tp, Tr);
T11 = VSUB(TZ, T10);
T1f = VADD(TZ, T10);
TD = VFNMSCONJ(Tj, Tg);
TE = VADD(Tu, Tv);
TF = VMUL(LDK(KP866025403), VSUB(TD, TE));
TP = VMUL(LDK(KP866025403), VADD(TE, TD));
To = VFNMS(LDK(KP500000000), VCONJ(Tn), Tm);
Ts = VFNMS(LDK(KP500000000), Tr, Tq);
Tt = VSUB(To, Ts);
TB = VADD(To, Ts);
TW = VADD(T2, T5);
TX = VFMACONJ(T7, Tb);
TY = VSUB(TW, TX);
T1e = VADD(TW, TX);
}
{
V T1l, T12, TG, TU, Ty, T1k, TV, TC, Tz, TT, Tl, Tx, T1, T1j, TH;
V TI, T1n, T1m, T14, T13, T18, T1g, TQ, T16, TM, T1c, T17, T1d, TO, TN;
V T15, TK, TL, TJ, T1b, TR, TS, T1i, T1h, T1a, T19;
T1l = VADD(T1e, T1f);
TV = LDW(&(W[TWVL * 4]));
T12 = VZMULI(TV, VFNMSI(T11, TY));
TC = VSUB(TA, TB);
Tz = LDW(&(W[TWVL * 18]));
TG = VZMUL(Tz, VFNMSI(TF, TC));
TT = LDW(&(W[TWVL * 2]));
TU = VZMUL(TT, VFMAI(TF, TC));
Tl = VFMA(LDK(KP866025403), Tk, Td);
Tx = VFMA(LDK(KP866025403), Tw, Tt);
T1 = LDW(&(W[TWVL * 20]));
Ty = VZMULI(T1, VFNMSI(Tx, Tl));
T1j = LDW(&(W[0]));
T1k = VZMULI(T1j, VFMAI(Tx, Tl));
TH = VADD(Ty, TG);
ST(&(Rp[WS(rs, 5)]), TH, ms, &(Rp[WS(rs, 1)]));
TI = VCONJ(VSUB(TG, Ty));
ST(&(Rm[WS(rs, 5)]), TI, -ms, &(Rm[WS(rs, 1)]));
T1n = VCONJ(VSUB(T1l, T1k));
ST(&(Rm[0]), T1n, -ms, &(Rm[0]));
T1m = VADD(T1k, T1l);
ST(&(Rp[0]), T1m, ms, &(Rp[0]));
T14 = VADD(TU, T12);
ST(&(Rp[WS(rs, 1)]), T14, ms, &(Rp[WS(rs, 1)]));
T13 = VCONJ(VSUB(TU, T12));
ST(&(Rm[WS(rs, 1)]), T13, -ms, &(Rm[WS(rs, 1)]));
T17 = LDW(&(W[TWVL * 16]));
T18 = VZMULI(T17, VFMAI(T11, TY));
T1d = LDW(&(W[TWVL * 10]));
T1g = VZMUL(T1d, VSUB(T1e, T1f));
TO = VADD(TA, TB);
TN = LDW(&(W[TWVL * 6]));
TQ = VZMUL(TN, VFMAI(TP, TO));
T15 = LDW(&(W[TWVL * 14]));
T16 = VZMUL(T15, VFNMSI(TP, TO));
TK = VFNMS(LDK(KP866025403), Tk, Td);
TL = VFNMS(LDK(KP866025403), Tw, Tt);
TJ = LDW(&(W[TWVL * 8]));
TM = VZMULI(TJ, VFMAI(TL, TK));
T1b = LDW(&(W[TWVL * 12]));
T1c = VZMULI(T1b, VFNMSI(TL, TK));
TR = VADD(TM, TQ);
ST(&(Rp[WS(rs, 2)]), TR, ms, &(Rp[0]));
TS = VCONJ(VSUB(TQ, TM));
ST(&(Rm[WS(rs, 2)]), TS, -ms, &(Rm[0]));
T1i = VCONJ(VSUB(T1g, T1c));
ST(&(Rm[WS(rs, 3)]), T1i, -ms, &(Rm[WS(rs, 1)]));
T1h = VADD(T1c, T1g);
ST(&(Rp[WS(rs, 3)]), T1h, ms, &(Rp[WS(rs, 1)]));
T1a = VADD(T16, T18);
ST(&(Rp[WS(rs, 4)]), T1a, ms, &(Rp[0]));
T19 = VCONJ(VSUB(T16, T18));
ST(&(Rm[WS(rs, 4)]), T19, -ms, &(Rm[0]));
}
}
}
VLEAVE();
}
static const tw_instr twinstr[] = {
VTW(1, 1),
VTW(1, 2),
VTW(1, 3),
VTW(1, 4),
VTW(1, 5),
VTW(1, 6),
VTW(1, 7),
VTW(1, 8),
VTW(1, 9),
VTW(1, 10),
VTW(1, 11),
{ TW_NEXT, VL, 0 }
};
static const hc2c_desc desc = { 12, XSIMD_STRING("hc2cbdftv_12"), twinstr, &GENUS, { 45, 25, 26, 0 } };
void XSIMD(codelet_hc2cbdftv_12) (planner *p) {
X(khc2c_register) (p, hc2cbdftv_12, &desc, HC2C_VIA_DFT);
}
#else
/* Generated by: ../../../genfft/gen_hc2cdft_c.native -simd -compact -variables 4 -pipeline-latency 8 -trivial-stores -variables 32 -no-generate-bytw -n 12 -dif -sign 1 -name hc2cbdftv_12 -include rdft/simd/hc2cbv.h */
/*
* This function contains 71 FP additions, 30 FP multiplications,
* (or, 67 additions, 26 multiplications, 4 fused multiply/add),
* 90 stack variables, 2 constants, and 24 memory accesses
*/
#include "rdft/simd/hc2cbv.h"
static void hc2cbdftv_12(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
{
INT m;
for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 22)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 22), MAKE_VOLATILE_STRIDE(48, rs)) {
V TY, TZ, Tf, TC, Tq, TG, Tm, TF, Ty, TD, T13, T1h, T2, T9, T3;
V T5, T6, Tc, Tb, Td, T8, T4, Ta, T7, Te, To, Tp, Tr, Tv, Ti;
V Ts, Tl, Tw, Tu, Tg, Th, Tj, Tk, Tt, Tx, T11, T12;
T2 = LD(&(Rp[0]), ms, &(Rp[0]));
T8 = LD(&(Rm[WS(rs, 5)]), -ms, &(Rm[WS(rs, 1)]));
T9 = VCONJ(T8);
T3 = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
T4 = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
T5 = VCONJ(T4);
T6 = VADD(T3, T5);
Tc = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
Ta = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
Tb = VCONJ(Ta);
Td = VADD(Tb, Tc);
TY = VADD(T2, T6);
TZ = VADD(T9, Td);
T7 = VFNMS(LDK(KP500000000), T6, T2);
Te = VFNMS(LDK(KP500000000), Td, T9);
Tf = VSUB(T7, Te);
TC = VADD(T7, Te);
To = VSUB(T3, T5);
Tp = VSUB(Tb, Tc);
Tq = VMUL(LDK(KP866025403), VSUB(To, Tp));
TG = VADD(To, Tp);
Tr = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
Tu = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
Tv = VCONJ(Tu);
Tg = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
Th = LD(&(Rm[0]), -ms, &(Rm[0]));
Ti = VCONJ(VSUB(Tg, Th));
Ts = VCONJ(VADD(Tg, Th));
Tj = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
Tk = LD(&(Rp[WS(rs, 5)]), ms, &(Rp[WS(rs, 1)]));
Tl = VSUB(Tj, Tk);
Tw = VADD(Tj, Tk);
Tm = VMUL(LDK(KP866025403), VSUB(Ti, Tl));
TF = VADD(Ti, Tl);
Tt = VFNMS(LDK(KP500000000), Ts, Tr);
Tx = VFNMS(LDK(KP500000000), Tw, Tv);
Ty = VSUB(Tt, Tx);
TD = VADD(Tt, Tx);
T11 = VADD(Tr, Ts);
T12 = VADD(Tv, Tw);
T13 = VBYI(VSUB(T11, T12));
T1h = VADD(T11, T12);
{
V T1n, T1i, T14, T1a, TA, T1m, TS, T18, TO, T1e, TI, TW, T1g, T1f, T10;
V TX, T19, Tn, Tz, T1, T1l, TQ, TR, TP, T17, TM, TN, TL, T1d, TE;
V TH, TB, TV, TJ, T1p, T1k, TT, T1o, TK, TU, T1j, T1b, T16, T1c, T15;
T1g = VADD(TY, TZ);
T1n = VADD(T1g, T1h);
T1f = LDW(&(W[TWVL * 10]));
T1i = VZMUL(T1f, VSUB(T1g, T1h));
T10 = VSUB(TY, TZ);
TX = LDW(&(W[TWVL * 4]));
T14 = VZMULI(TX, VSUB(T10, T13));
T19 = LDW(&(W[TWVL * 16]));
T1a = VZMULI(T19, VADD(T10, T13));
Tn = VSUB(Tf, Tm);
Tz = VBYI(VADD(Tq, Ty));
T1 = LDW(&(W[TWVL * 20]));
TA = VZMULI(T1, VSUB(Tn, Tz));
T1l = LDW(&(W[0]));
T1m = VZMULI(T1l, VADD(Tn, Tz));
TQ = VBYI(VMUL(LDK(KP866025403), VADD(TG, TF)));
TR = VADD(TC, TD);
TP = LDW(&(W[TWVL * 6]));
TS = VZMUL(TP, VADD(TQ, TR));
T17 = LDW(&(W[TWVL * 14]));
T18 = VZMUL(T17, VSUB(TR, TQ));
TM = VADD(Tf, Tm);
TN = VBYI(VSUB(Ty, Tq));
TL = LDW(&(W[TWVL * 8]));
TO = VZMULI(TL, VADD(TM, TN));
T1d = LDW(&(W[TWVL * 12]));
T1e = VZMULI(T1d, VSUB(TM, TN));
TE = VSUB(TC, TD);
TH = VBYI(VMUL(LDK(KP866025403), VSUB(TF, TG)));
TB = LDW(&(W[TWVL * 18]));
TI = VZMUL(TB, VSUB(TE, TH));
TV = LDW(&(W[TWVL * 2]));
TW = VZMUL(TV, VADD(TH, TE));
TJ = VADD(TA, TI);
ST(&(Rp[WS(rs, 5)]), TJ, ms, &(Rp[WS(rs, 1)]));
T1p = VCONJ(VSUB(T1n, T1m));
ST(&(Rm[0]), T1p, -ms, &(Rm[0]));
T1k = VCONJ(VSUB(T1i, T1e));
ST(&(Rm[WS(rs, 3)]), T1k, -ms, &(Rm[WS(rs, 1)]));
TT = VADD(TO, TS);
ST(&(Rp[WS(rs, 2)]), TT, ms, &(Rp[0]));
T1o = VADD(T1m, T1n);
ST(&(Rp[0]), T1o, ms, &(Rp[0]));
TK = VCONJ(VSUB(TI, TA));
ST(&(Rm[WS(rs, 5)]), TK, -ms, &(Rm[WS(rs, 1)]));
TU = VCONJ(VSUB(TS, TO));
ST(&(Rm[WS(rs, 2)]), TU, -ms, &(Rm[0]));
T1j = VADD(T1e, T1i);
ST(&(Rp[WS(rs, 3)]), T1j, ms, &(Rp[WS(rs, 1)]));
T1b = VCONJ(VSUB(T18, T1a));
ST(&(Rm[WS(rs, 4)]), T1b, -ms, &(Rm[0]));
T16 = VADD(TW, T14);
ST(&(Rp[WS(rs, 1)]), T16, ms, &(Rp[WS(rs, 1)]));
T1c = VADD(T18, T1a);
ST(&(Rp[WS(rs, 4)]), T1c, ms, &(Rp[0]));
T15 = VCONJ(VSUB(TW, T14));
ST(&(Rm[WS(rs, 1)]), T15, -ms, &(Rm[WS(rs, 1)]));
}
}
}
VLEAVE();
}
static const tw_instr twinstr[] = {
VTW(1, 1),
VTW(1, 2),
VTW(1, 3),
VTW(1, 4),
VTW(1, 5),
VTW(1, 6),
VTW(1, 7),
VTW(1, 8),
VTW(1, 9),
VTW(1, 10),
VTW(1, 11),
{ TW_NEXT, VL, 0 }
};
static const hc2c_desc desc = { 12, XSIMD_STRING("hc2cbdftv_12"), twinstr, &GENUS, { 67, 26, 4, 0 } };
void XSIMD(codelet_hc2cbdftv_12) (planner *p) {
X(khc2c_register) (p, hc2cbdftv_12, &desc, HC2C_VIA_DFT);
}
#endif
Reference in a new issue View git blame Copy permalink