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 07:39:02 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_twiddle_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 3 -name t1buv_3 -include t1bu.h -sign 1 */
Chris@10: 
Chris@10: /*
Chris@10:  * This function contains 8 FP additions, 8 FP multiplications,
Chris@10:  * (or, 5 additions, 5 multiplications, 3 fused multiply/add),
Chris@10:  * 12 stack variables, 2 constants, and 6 memory accesses
Chris@10:  */
Chris@10: #include "t1bu.h"
Chris@10: 
Chris@10: static void t1buv_3(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@10: {
Chris@10:      DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@10:      DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@10:      {
Chris@10: 	  INT m;
Chris@10: 	  R *x;
Chris@10: 	  x = ii;
Chris@10: 	  for (m = mb, W = W + (mb * ((TWVL / VL) * 4)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 4), MAKE_VOLATILE_STRIDE(3, rs)) {
Chris@10: 	       V T1, T2, T4;
Chris@10: 	       T1 = LD(&(x[0]), ms, &(x[0]));
Chris@10: 	       T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@10: 	       T4 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@10: 	       {
Chris@10: 		    V T3, T5, T8, T6, T7;
Chris@10: 		    T3 = BYTW(&(W[0]), T2);
Chris@10: 		    T5 = BYTW(&(W[TWVL * 2]), T4);
Chris@10: 		    T8 = VMUL(LDK(KP866025403), VSUB(T3, T5));
Chris@10: 		    T6 = VADD(T3, T5);
Chris@10: 		    T7 = VFNMS(LDK(KP500000000), T6, T1);
Chris@10: 		    ST(&(x[0]), VADD(T1, T6), ms, &(x[0]));
Chris@10: 		    ST(&(x[WS(rs, 2)]), VFNMSI(T8, T7), ms, &(x[0]));
Chris@10: 		    ST(&(x[WS(rs, 1)]), VFMAI(T8, T7), ms, &(x[WS(rs, 1)]));
Chris@10: 	       }
Chris@10: 	  }
Chris@10:      }
Chris@10:      VLEAVE();
Chris@10: }
Chris@10: 
Chris@10: static const tw_instr twinstr[] = {
Chris@10:      VTW(0, 1),
Chris@10:      VTW(0, 2),
Chris@10:      {TW_NEXT, VL, 0}
Chris@10: };
Chris@10: 
Chris@10: static const ct_desc desc = { 3, XSIMD_STRING("t1buv_3"), twinstr, &GENUS, {5, 5, 3, 0}, 0, 0, 0 };
Chris@10: 
Chris@10: void XSIMD(codelet_t1buv_3) (planner *p) {
Chris@10:      X(kdft_dit_register) (p, t1buv_3, &desc);
Chris@10: }
Chris@10: #else				/* HAVE_FMA */
Chris@10: 
Chris@10: /* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 3 -name t1buv_3 -include t1bu.h -sign 1 */
Chris@10: 
Chris@10: /*
Chris@10:  * This function contains 8 FP additions, 6 FP multiplications,
Chris@10:  * (or, 7 additions, 5 multiplications, 1 fused multiply/add),
Chris@10:  * 12 stack variables, 2 constants, and 6 memory accesses
Chris@10:  */
Chris@10: #include "t1bu.h"
Chris@10: 
Chris@10: static void t1buv_3(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@10: {
Chris@10:      DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@10:      DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@10:      {
Chris@10: 	  INT m;
Chris@10: 	  R *x;
Chris@10: 	  x = ii;
Chris@10: 	  for (m = mb, W = W + (mb * ((TWVL / VL) * 4)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 4), MAKE_VOLATILE_STRIDE(3, rs)) {
Chris@10: 	       V T6, T2, T4, T7, T1, T3, T5, T8;
Chris@10: 	       T6 = LD(&(x[0]), ms, &(x[0]));
Chris@10: 	       T1 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@10: 	       T2 = BYTW(&(W[0]), T1);
Chris@10: 	       T3 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@10: 	       T4 = BYTW(&(W[TWVL * 2]), T3);
Chris@10: 	       T7 = VADD(T2, T4);
Chris@10: 	       ST(&(x[0]), VADD(T6, T7), ms, &(x[0]));
Chris@10: 	       T5 = VBYI(VMUL(LDK(KP866025403), VSUB(T2, T4)));
Chris@10: 	       T8 = VFNMS(LDK(KP500000000), T7, T6);
Chris@10: 	       ST(&(x[WS(rs, 1)]), VADD(T5, T8), ms, &(x[WS(rs, 1)]));
Chris@10: 	       ST(&(x[WS(rs, 2)]), VSUB(T8, T5), ms, &(x[0]));
Chris@10: 	  }
Chris@10:      }
Chris@10:      VLEAVE();
Chris@10: }
Chris@10: 
Chris@10: static const tw_instr twinstr[] = {
Chris@10:      VTW(0, 1),
Chris@10:      VTW(0, 2),
Chris@10:      {TW_NEXT, VL, 0}
Chris@10: };
Chris@10: 
Chris@10: static const ct_desc desc = { 3, XSIMD_STRING("t1buv_3"), twinstr, &GENUS, {7, 5, 1, 0}, 0, 0, 0 };
Chris@10: 
Chris@10: void XSIMD(codelet_t1buv_3) (planner *p) {
Chris@10:      X(kdft_dit_register) (p, t1buv_3, &desc);
Chris@10: }
Chris@10: #endif				/* HAVE_FMA */