Chris@10: /*
Chris@10:  * Copyright (c) 2003, 2007-11 Matteo Frigo
Chris@10:  * Copyright (c) 2003, 2007-11 Massachusetts Institute of Technology
Chris@10:  *
Chris@10:  * This program is free software; you can redistribute it and/or modify
Chris@10:  * it under the terms of the GNU General Public License as published by
Chris@10:  * the Free Software Foundation; either version 2 of the License, or
Chris@10:  * (at your option) any later version.
Chris@10:  *
Chris@10:  * This program is distributed in the hope that it will be useful,
Chris@10:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@10:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
Chris@10:  * GNU General Public License for more details.
Chris@10:  *
Chris@10:  * You should have received a copy of the GNU General Public License
Chris@10:  * along with this program; if not, write to the Free Software
Chris@10:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
Chris@10:  *
Chris@10:  */
Chris@10: 
Chris@10: /* This file was automatically generated --- DO NOT EDIT */
Chris@10: /* Generated on Sun Nov 25 07:36:58 EST 2012 */
Chris@10: 
Chris@10: #include "codelet-dft.h"
Chris@10: 
Chris@10: #ifdef HAVE_FMA
Chris@10: 
Chris@10: /* Generated by: ../../../genfft/gen_notw_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 3 -name n1bv_3 -include n1b.h */
Chris@10: 
Chris@10: /*
Chris@10:  * This function contains 6 FP additions, 4 FP multiplications,
Chris@10:  * (or, 3 additions, 1 multiplications, 3 fused multiply/add),
Chris@10:  * 11 stack variables, 2 constants, and 6 memory accesses
Chris@10:  */
Chris@10: #include "n1b.h"
Chris@10: 
Chris@10: static void n1bv_3(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@10: {
Chris@10:      DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@10:      DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@10:      {
Chris@10: 	  INT i;
Chris@10: 	  const R *xi;
Chris@10: 	  R *xo;
Chris@10: 	  xi = ii;
Chris@10: 	  xo = io;
Chris@10: 	  for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(6, is), MAKE_VOLATILE_STRIDE(6, os)) {
Chris@10: 	       V T1, T2, T3, T6, T4, T5;
Chris@10: 	       T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@10: 	       T2 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@10: 	       T3 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@10: 	       T6 = VMUL(LDK(KP866025403), VSUB(T2, T3));
Chris@10: 	       T4 = VADD(T2, T3);
Chris@10: 	       T5 = VFNMS(LDK(KP500000000), T4, T1);
Chris@10: 	       ST(&(xo[0]), VADD(T1, T4), ovs, &(xo[0]));
Chris@10: 	       ST(&(xo[WS(os, 2)]), VFNMSI(T6, T5), ovs, &(xo[0]));
Chris@10: 	       ST(&(xo[WS(os, 1)]), VFMAI(T6, T5), ovs, &(xo[WS(os, 1)]));
Chris@10: 	  }
Chris@10:      }
Chris@10:      VLEAVE();
Chris@10: }
Chris@10: 
Chris@10: static const kdft_desc desc = { 3, XSIMD_STRING("n1bv_3"), {3, 1, 3, 0}, &GENUS, 0, 0, 0, 0 };
Chris@10: 
Chris@10: void XSIMD(codelet_n1bv_3) (planner *p) {
Chris@10:      X(kdft_register) (p, n1bv_3, &desc);
Chris@10: }
Chris@10: 
Chris@10: #else				/* HAVE_FMA */
Chris@10: 
Chris@10: /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 3 -name n1bv_3 -include n1b.h */
Chris@10: 
Chris@10: /*
Chris@10:  * This function contains 6 FP additions, 2 FP multiplications,
Chris@10:  * (or, 5 additions, 1 multiplications, 1 fused multiply/add),
Chris@10:  * 11 stack variables, 2 constants, and 6 memory accesses
Chris@10:  */
Chris@10: #include "n1b.h"
Chris@10: 
Chris@10: static void n1bv_3(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@10: {
Chris@10:      DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@10:      DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@10:      {
Chris@10: 	  INT i;
Chris@10: 	  const R *xi;
Chris@10: 	  R *xo;
Chris@10: 	  xi = ii;
Chris@10: 	  xo = io;
Chris@10: 	  for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(6, is), MAKE_VOLATILE_STRIDE(6, os)) {
Chris@10: 	       V T4, T3, T5, T1, T2, T6;
Chris@10: 	       T4 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@10: 	       T1 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@10: 	       T2 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@10: 	       T3 = VBYI(VMUL(LDK(KP866025403), VSUB(T1, T2)));
Chris@10: 	       T5 = VADD(T1, T2);
Chris@10: 	       ST(&(xo[0]), VADD(T4, T5), ovs, &(xo[0]));
Chris@10: 	       T6 = VFNMS(LDK(KP500000000), T5, T4);
Chris@10: 	       ST(&(xo[WS(os, 1)]), VADD(T3, T6), ovs, &(xo[WS(os, 1)]));
Chris@10: 	       ST(&(xo[WS(os, 2)]), VSUB(T6, T3), ovs, &(xo[0]));
Chris@10: 	  }
Chris@10:      }
Chris@10:      VLEAVE();
Chris@10: }
Chris@10: 
Chris@10: static const kdft_desc desc = { 3, XSIMD_STRING("n1bv_3"), {5, 1, 1, 0}, &GENUS, 0, 0, 0, 0 };
Chris@10: 
Chris@10: void XSIMD(codelet_n1bv_3) (planner *p) {
Chris@10:      X(kdft_register) (p, n1bv_3, &desc);
Chris@10: }
Chris@10: 
Chris@10: #endif				/* HAVE_FMA */