cannam@167: /*
cannam@167:  * Copyright (c) 2003, 2007-14 Matteo Frigo
cannam@167:  * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
cannam@167:  *
cannam@167:  * This program is free software; you can redistribute it and/or modify
cannam@167:  * it under the terms of the GNU General Public License as published by
cannam@167:  * the Free Software Foundation; either version 2 of the License, or
cannam@167:  * (at your option) any later version.
cannam@167:  *
cannam@167:  * This program is distributed in the hope that it will be useful,
cannam@167:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
cannam@167:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
cannam@167:  * GNU General Public License for more details.
cannam@167:  *
cannam@167:  * You should have received a copy of the GNU General Public License
cannam@167:  * along with this program; if not, write to the Free Software
cannam@167:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
cannam@167:  *
cannam@167:  */
cannam@167: 
cannam@167: /* This file was automatically generated --- DO NOT EDIT */
cannam@167: /* Generated on Thu May 24 08:05:46 EDT 2018 */
cannam@167: 
cannam@167: #include "dft/codelet-dft.h"
cannam@167: 
cannam@167: #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
cannam@167: 
cannam@167: /* Generated by: ../../../genfft/gen_twiddle_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -n 5 -name t2fv_5 -include dft/simd/t2f.h */
cannam@167: 
cannam@167: /*
cannam@167:  * This function contains 20 FP additions, 19 FP multiplications,
cannam@167:  * (or, 11 additions, 10 multiplications, 9 fused multiply/add),
cannam@167:  * 20 stack variables, 4 constants, and 10 memory accesses
cannam@167:  */
cannam@167: #include "dft/simd/t2f.h"
cannam@167: 
cannam@167: static void t2fv_5(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
cannam@167: {
cannam@167:      DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
cannam@167:      DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
cannam@167:      DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
cannam@167:      DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
cannam@167:      {
cannam@167: 	  INT m;
cannam@167: 	  R *x;
cannam@167: 	  x = ri;
cannam@167: 	  for (m = mb, W = W + (mb * ((TWVL / VL) * 8)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 8), MAKE_VOLATILE_STRIDE(5, rs)) {
cannam@167: 	       V T1, Tg, Th, T6, Tb, Tc;
cannam@167: 	       T1 = LD(&(x[0]), ms, &(x[0]));
cannam@167: 	       {
cannam@167: 		    V T3, Ta, T5, T8;
cannam@167: 		    {
cannam@167: 			 V T2, T9, T4, T7;
cannam@167: 			 T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
cannam@167: 			 T3 = BYTWJ(&(W[0]), T2);
cannam@167: 			 T9 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
cannam@167: 			 Ta = BYTWJ(&(W[TWVL * 4]), T9);
cannam@167: 			 T4 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
cannam@167: 			 T5 = BYTWJ(&(W[TWVL * 6]), T4);
cannam@167: 			 T7 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
cannam@167: 			 T8 = BYTWJ(&(W[TWVL * 2]), T7);
cannam@167: 		    }
cannam@167: 		    Tg = VSUB(T3, T5);
cannam@167: 		    Th = VSUB(T8, Ta);
cannam@167: 		    T6 = VADD(T3, T5);
cannam@167: 		    Tb = VADD(T8, Ta);
cannam@167: 		    Tc = VADD(T6, Tb);
cannam@167: 	       }
cannam@167: 	       ST(&(x[0]), VADD(T1, Tc), ms, &(x[0]));
cannam@167: 	       {
cannam@167: 		    V Ti, Tk, Tf, Tj, Td, Te;
cannam@167: 		    Ti = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), Th, Tg));
cannam@167: 		    Tk = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), Tg, Th));
cannam@167: 		    Td = VFNMS(LDK(KP250000000), Tc, T1);
cannam@167: 		    Te = VSUB(T6, Tb);
cannam@167: 		    Tf = VFMA(LDK(KP559016994), Te, Td);
cannam@167: 		    Tj = VFNMS(LDK(KP559016994), Te, Td);
cannam@167: 		    ST(&(x[WS(rs, 1)]), VFNMSI(Ti, Tf), ms, &(x[WS(rs, 1)]));
cannam@167: 		    ST(&(x[WS(rs, 3)]), VFNMSI(Tk, Tj), ms, &(x[WS(rs, 1)]));
cannam@167: 		    ST(&(x[WS(rs, 4)]), VFMAI(Ti, Tf), ms, &(x[0]));
cannam@167: 		    ST(&(x[WS(rs, 2)]), VFMAI(Tk, Tj), ms, &(x[0]));
cannam@167: 	       }
cannam@167: 	  }
cannam@167:      }
cannam@167:      VLEAVE();
cannam@167: }
cannam@167: 
cannam@167: static const tw_instr twinstr[] = {
cannam@167:      VTW(0, 1),
cannam@167:      VTW(0, 2),
cannam@167:      VTW(0, 3),
cannam@167:      VTW(0, 4),
cannam@167:      {TW_NEXT, VL, 0}
cannam@167: };
cannam@167: 
cannam@167: static const ct_desc desc = { 5, XSIMD_STRING("t2fv_5"), twinstr, &GENUS, {11, 10, 9, 0}, 0, 0, 0 };
cannam@167: 
cannam@167: void XSIMD(codelet_t2fv_5) (planner *p) {
cannam@167:      X(kdft_dit_register) (p, t2fv_5, &desc);
cannam@167: }
cannam@167: #else
cannam@167: 
cannam@167: /* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 5 -name t2fv_5 -include dft/simd/t2f.h */
cannam@167: 
cannam@167: /*
cannam@167:  * This function contains 20 FP additions, 14 FP multiplications,
cannam@167:  * (or, 17 additions, 11 multiplications, 3 fused multiply/add),
cannam@167:  * 20 stack variables, 4 constants, and 10 memory accesses
cannam@167:  */
cannam@167: #include "dft/simd/t2f.h"
cannam@167: 
cannam@167: static void t2fv_5(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
cannam@167: {
cannam@167:      DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
cannam@167:      DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
cannam@167:      DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
cannam@167:      DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
cannam@167:      {
cannam@167: 	  INT m;
cannam@167: 	  R *x;
cannam@167: 	  x = ri;
cannam@167: 	  for (m = mb, W = W + (mb * ((TWVL / VL) * 8)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 8), MAKE_VOLATILE_STRIDE(5, rs)) {
cannam@167: 	       V Tc, Tg, Th, T5, Ta, Td;
cannam@167: 	       Tc = LD(&(x[0]), ms, &(x[0]));
cannam@167: 	       {
cannam@167: 		    V T2, T9, T4, T7;
cannam@167: 		    {
cannam@167: 			 V T1, T8, T3, T6;
cannam@167: 			 T1 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
cannam@167: 			 T2 = BYTWJ(&(W[0]), T1);
cannam@167: 			 T8 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
cannam@167: 			 T9 = BYTWJ(&(W[TWVL * 4]), T8);
cannam@167: 			 T3 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
cannam@167: 			 T4 = BYTWJ(&(W[TWVL * 6]), T3);
cannam@167: 			 T6 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
cannam@167: 			 T7 = BYTWJ(&(W[TWVL * 2]), T6);
cannam@167: 		    }
cannam@167: 		    Tg = VSUB(T2, T4);
cannam@167: 		    Th = VSUB(T7, T9);
cannam@167: 		    T5 = VADD(T2, T4);
cannam@167: 		    Ta = VADD(T7, T9);
cannam@167: 		    Td = VADD(T5, Ta);
cannam@167: 	       }
cannam@167: 	       ST(&(x[0]), VADD(Tc, Td), ms, &(x[0]));
cannam@167: 	       {
cannam@167: 		    V Ti, Tj, Tf, Tk, Tb, Te;
cannam@167: 		    Ti = VBYI(VFMA(LDK(KP951056516), Tg, VMUL(LDK(KP587785252), Th)));
cannam@167: 		    Tj = VBYI(VFNMS(LDK(KP587785252), Tg, VMUL(LDK(KP951056516), Th)));
cannam@167: 		    Tb = VMUL(LDK(KP559016994), VSUB(T5, Ta));
cannam@167: 		    Te = VFNMS(LDK(KP250000000), Td, Tc);
cannam@167: 		    Tf = VADD(Tb, Te);
cannam@167: 		    Tk = VSUB(Te, Tb);
cannam@167: 		    ST(&(x[WS(rs, 1)]), VSUB(Tf, Ti), ms, &(x[WS(rs, 1)]));
cannam@167: 		    ST(&(x[WS(rs, 3)]), VSUB(Tk, Tj), ms, &(x[WS(rs, 1)]));
cannam@167: 		    ST(&(x[WS(rs, 4)]), VADD(Ti, Tf), ms, &(x[0]));
cannam@167: 		    ST(&(x[WS(rs, 2)]), VADD(Tj, Tk), ms, &(x[0]));
cannam@167: 	       }
cannam@167: 	  }
cannam@167:      }
cannam@167:      VLEAVE();
cannam@167: }
cannam@167: 
cannam@167: static const tw_instr twinstr[] = {
cannam@167:      VTW(0, 1),
cannam@167:      VTW(0, 2),
cannam@167:      VTW(0, 3),
cannam@167:      VTW(0, 4),
cannam@167:      {TW_NEXT, VL, 0}
cannam@167: };
cannam@167: 
cannam@167: static const ct_desc desc = { 5, XSIMD_STRING("t2fv_5"), twinstr, &GENUS, {17, 11, 3, 0}, 0, 0, 0 };
cannam@167: 
cannam@167: void XSIMD(codelet_t2fv_5) (planner *p) {
cannam@167:      X(kdft_dit_register) (p, t2fv_5, &desc);
cannam@167: }
cannam@167: #endif