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 08:05:11 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_notw_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 8 -name n2bv_8 -with-ostride 2 -include dft/simd/n2b.h -store-multiple 2 */
cannam@167: 
cannam@167: /*
cannam@167:  * This function contains 26 FP additions, 10 FP multiplications,
cannam@167:  * (or, 16 additions, 0 multiplications, 10 fused multiply/add),
cannam@167:  * 24 stack variables, 1 constants, and 20 memory accesses
cannam@167:  */
cannam@167: #include "dft/simd/n2b.h"
cannam@167: 
cannam@167: static void n2bv_8(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
cannam@167: {
cannam@167:      DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
cannam@167:      {
cannam@167: 	  INT i;
cannam@167: 	  const R *xi;
cannam@167: 	  R *xo;
cannam@167: 	  xi = ii;
cannam@167: 	  xo = io;
cannam@167: 	  for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(16, is), MAKE_VOLATILE_STRIDE(16, os)) {
cannam@167: 	       V T3, Tj, Te, Tk, Ta, Tn, Tf, Tm, Tr, Tu;
cannam@167: 	       {
cannam@167: 		    V T1, T2, Tc, Td;
cannam@167: 		    T1 = LD(&(xi[0]), ivs, &(xi[0]));
cannam@167: 		    T2 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
cannam@167: 		    T3 = VSUB(T1, T2);
cannam@167: 		    Tj = VADD(T1, T2);
cannam@167: 		    Tc = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
cannam@167: 		    Td = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
cannam@167: 		    Te = VSUB(Tc, Td);
cannam@167: 		    Tk = VADD(Tc, Td);
cannam@167: 		    {
cannam@167: 			 V T4, T5, T6, T7, T8, T9;
cannam@167: 			 T4 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
cannam@167: 			 T5 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
cannam@167: 			 T6 = VSUB(T4, T5);
cannam@167: 			 T7 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
cannam@167: 			 T8 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
cannam@167: 			 T9 = VSUB(T7, T8);
cannam@167: 			 Ta = VADD(T6, T9);
cannam@167: 			 Tn = VADD(T7, T8);
cannam@167: 			 Tf = VSUB(T6, T9);
cannam@167: 			 Tm = VADD(T4, T5);
cannam@167: 		    }
cannam@167: 	       }
cannam@167: 	       {
cannam@167: 		    V Ts, Tb, Tg, Tp, Tq, Tt;
cannam@167: 		    Tb = VFNMS(LDK(KP707106781), Ta, T3);
cannam@167: 		    Tg = VFNMS(LDK(KP707106781), Tf, Te);
cannam@167: 		    Tr = VFNMSI(Tg, Tb);
cannam@167: 		    STM2(&(xo[6]), Tr, ovs, &(xo[2]));
cannam@167: 		    Ts = VFMAI(Tg, Tb);
cannam@167: 		    STM2(&(xo[10]), Ts, ovs, &(xo[2]));
cannam@167: 		    Tp = VADD(Tj, Tk);
cannam@167: 		    Tq = VADD(Tm, Tn);
cannam@167: 		    Tt = VSUB(Tp, Tq);
cannam@167: 		    STM2(&(xo[8]), Tt, ovs, &(xo[0]));
cannam@167: 		    STN2(&(xo[8]), Tt, Ts, ovs);
cannam@167: 		    Tu = VADD(Tp, Tq);
cannam@167: 		    STM2(&(xo[0]), Tu, ovs, &(xo[0]));
cannam@167: 	       }
cannam@167: 	       {
cannam@167: 		    V Tw, Th, Ti, Tv;
cannam@167: 		    Th = VFMA(LDK(KP707106781), Ta, T3);
cannam@167: 		    Ti = VFMA(LDK(KP707106781), Tf, Te);
cannam@167: 		    Tv = VFMAI(Ti, Th);
cannam@167: 		    STM2(&(xo[2]), Tv, ovs, &(xo[2]));
cannam@167: 		    STN2(&(xo[0]), Tu, Tv, ovs);
cannam@167: 		    Tw = VFNMSI(Ti, Th);
cannam@167: 		    STM2(&(xo[14]), Tw, ovs, &(xo[2]));
cannam@167: 		    {
cannam@167: 			 V Tl, To, Tx, Ty;
cannam@167: 			 Tl = VSUB(Tj, Tk);
cannam@167: 			 To = VSUB(Tm, Tn);
cannam@167: 			 Tx = VFNMSI(To, Tl);
cannam@167: 			 STM2(&(xo[12]), Tx, ovs, &(xo[0]));
cannam@167: 			 STN2(&(xo[12]), Tx, Tw, ovs);
cannam@167: 			 Ty = VFMAI(To, Tl);
cannam@167: 			 STM2(&(xo[4]), Ty, ovs, &(xo[0]));
cannam@167: 			 STN2(&(xo[4]), Ty, Tr, ovs);
cannam@167: 		    }
cannam@167: 	       }
cannam@167: 	  }
cannam@167:      }
cannam@167:      VLEAVE();
cannam@167: }
cannam@167: 
cannam@167: static const kdft_desc desc = { 8, XSIMD_STRING("n2bv_8"), {16, 0, 10, 0}, &GENUS, 0, 2, 0, 0 };
cannam@167: 
cannam@167: void XSIMD(codelet_n2bv_8) (planner *p) {
cannam@167:      X(kdft_register) (p, n2bv_8, &desc);
cannam@167: }
cannam@167: 
cannam@167: #else
cannam@167: 
cannam@167: /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 8 -name n2bv_8 -with-ostride 2 -include dft/simd/n2b.h -store-multiple 2 */
cannam@167: 
cannam@167: /*
cannam@167:  * This function contains 26 FP additions, 2 FP multiplications,
cannam@167:  * (or, 26 additions, 2 multiplications, 0 fused multiply/add),
cannam@167:  * 24 stack variables, 1 constants, and 20 memory accesses
cannam@167:  */
cannam@167: #include "dft/simd/n2b.h"
cannam@167: 
cannam@167: static void n2bv_8(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
cannam@167: {
cannam@167:      DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
cannam@167:      {
cannam@167: 	  INT i;
cannam@167: 	  const R *xi;
cannam@167: 	  R *xo;
cannam@167: 	  xi = ii;
cannam@167: 	  xo = io;
cannam@167: 	  for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(16, is), MAKE_VOLATILE_STRIDE(16, os)) {
cannam@167: 	       V Ta, Tk, Te, Tj, T7, Tn, Tf, Tm, Tr, Tu;
cannam@167: 	       {
cannam@167: 		    V T8, T9, Tc, Td;
cannam@167: 		    T8 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
cannam@167: 		    T9 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
cannam@167: 		    Ta = VSUB(T8, T9);
cannam@167: 		    Tk = VADD(T8, T9);
cannam@167: 		    Tc = LD(&(xi[0]), ivs, &(xi[0]));
cannam@167: 		    Td = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
cannam@167: 		    Te = VSUB(Tc, Td);
cannam@167: 		    Tj = VADD(Tc, Td);
cannam@167: 		    {
cannam@167: 			 V T1, T2, T3, T4, T5, T6;
cannam@167: 			 T1 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
cannam@167: 			 T2 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
cannam@167: 			 T3 = VSUB(T1, T2);
cannam@167: 			 T4 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
cannam@167: 			 T5 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
cannam@167: 			 T6 = VSUB(T4, T5);
cannam@167: 			 T7 = VMUL(LDK(KP707106781), VSUB(T3, T6));
cannam@167: 			 Tn = VADD(T4, T5);
cannam@167: 			 Tf = VMUL(LDK(KP707106781), VADD(T3, T6));
cannam@167: 			 Tm = VADD(T1, T2);
cannam@167: 		    }
cannam@167: 	       }
cannam@167: 	       {
cannam@167: 		    V Ts, Tb, Tg, Tp, Tq, Tt;
cannam@167: 		    Tb = VBYI(VSUB(T7, Ta));
cannam@167: 		    Tg = VSUB(Te, Tf);
cannam@167: 		    Tr = VADD(Tb, Tg);
cannam@167: 		    STM2(&(xo[6]), Tr, ovs, &(xo[2]));
cannam@167: 		    Ts = VSUB(Tg, Tb);
cannam@167: 		    STM2(&(xo[10]), Ts, ovs, &(xo[2]));
cannam@167: 		    Tp = VADD(Tj, Tk);
cannam@167: 		    Tq = VADD(Tm, Tn);
cannam@167: 		    Tt = VSUB(Tp, Tq);
cannam@167: 		    STM2(&(xo[8]), Tt, ovs, &(xo[0]));
cannam@167: 		    STN2(&(xo[8]), Tt, Ts, ovs);
cannam@167: 		    Tu = VADD(Tp, Tq);
cannam@167: 		    STM2(&(xo[0]), Tu, ovs, &(xo[0]));
cannam@167: 	       }
cannam@167: 	       {
cannam@167: 		    V Tw, Th, Ti, Tv;
cannam@167: 		    Th = VBYI(VADD(Ta, T7));
cannam@167: 		    Ti = VADD(Te, Tf);
cannam@167: 		    Tv = VADD(Th, Ti);
cannam@167: 		    STM2(&(xo[2]), Tv, ovs, &(xo[2]));
cannam@167: 		    STN2(&(xo[0]), Tu, Tv, ovs);
cannam@167: 		    Tw = VSUB(Ti, Th);
cannam@167: 		    STM2(&(xo[14]), Tw, ovs, &(xo[2]));
cannam@167: 		    {
cannam@167: 			 V Tl, To, Tx, Ty;
cannam@167: 			 Tl = VSUB(Tj, Tk);
cannam@167: 			 To = VBYI(VSUB(Tm, Tn));
cannam@167: 			 Tx = VSUB(Tl, To);
cannam@167: 			 STM2(&(xo[12]), Tx, ovs, &(xo[0]));
cannam@167: 			 STN2(&(xo[12]), Tx, Tw, ovs);
cannam@167: 			 Ty = VADD(Tl, To);
cannam@167: 			 STM2(&(xo[4]), Ty, ovs, &(xo[0]));
cannam@167: 			 STN2(&(xo[4]), Ty, Tr, ovs);
cannam@167: 		    }
cannam@167: 	       }
cannam@167: 	  }
cannam@167:      }
cannam@167:      VLEAVE();
cannam@167: }
cannam@167: 
cannam@167: static const kdft_desc desc = { 8, XSIMD_STRING("n2bv_8"), {26, 2, 0, 0}, &GENUS, 0, 2, 0, 0 };
cannam@167: 
cannam@167: void XSIMD(codelet_n2bv_8) (planner *p) {
cannam@167:      X(kdft_register) (p, n2bv_8, &desc);
cannam@167: }
cannam@167: 
cannam@167: #endif