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 07:39:23 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.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 2 -name t1sv_2 -include ts.h */
Chris@10: 
Chris@10: /*
Chris@10:  * This function contains 6 FP additions, 4 FP multiplications,
Chris@10:  * (or, 4 additions, 2 multiplications, 2 fused multiply/add),
Chris@10:  * 11 stack variables, 0 constants, and 8 memory accesses
Chris@10:  */
Chris@10: #include "ts.h"
Chris@10: 
Chris@10: static void t1sv_2(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@10: {
Chris@10:      {
Chris@10: 	  INT m;
Chris@10: 	  for (m = mb, W = W + (mb * 2); m < me; m = m + (2 * VL), ri = ri + ((2 * VL) * ms), ii = ii + ((2 * VL) * ms), W = W + ((2 * VL) * 2), MAKE_VOLATILE_STRIDE(4, rs)) {
Chris@10: 	       V T1, Ta, T3, T6, T2, T5;
Chris@10: 	       T1 = LD(&(ri[0]), ms, &(ri[0]));
Chris@10: 	       Ta = LD(&(ii[0]), ms, &(ii[0]));
Chris@10: 	       T3 = LD(&(ri[WS(rs, 1)]), ms, &(ri[WS(rs, 1)]));
Chris@10: 	       T6 = LD(&(ii[WS(rs, 1)]), ms, &(ii[WS(rs, 1)]));
Chris@10: 	       T2 = LDW(&(W[0]));
Chris@10: 	       T5 = LDW(&(W[TWVL * 1]));
Chris@10: 	       {
Chris@10: 		    V T8, T4, T9, T7;
Chris@10: 		    T8 = VMUL(T2, T6);
Chris@10: 		    T4 = VMUL(T2, T3);
Chris@10: 		    T9 = VFNMS(T5, T3, T8);
Chris@10: 		    T7 = VFMA(T5, T6, T4);
Chris@10: 		    ST(&(ii[0]), VADD(T9, Ta), ms, &(ii[0]));
Chris@10: 		    ST(&(ii[WS(rs, 1)]), VSUB(Ta, T9), ms, &(ii[WS(rs, 1)]));
Chris@10: 		    ST(&(ri[0]), VADD(T1, T7), ms, &(ri[0]));
Chris@10: 		    ST(&(ri[WS(rs, 1)]), VSUB(T1, T7), ms, &(ri[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:      {TW_NEXT, (2 * VL), 0}
Chris@10: };
Chris@10: 
Chris@10: static const ct_desc desc = { 2, XSIMD_STRING("t1sv_2"), twinstr, &GENUS, {4, 2, 2, 0}, 0, 0, 0 };
Chris@10: 
Chris@10: void XSIMD(codelet_t1sv_2) (planner *p) {
Chris@10:      X(kdft_dit_register) (p, t1sv_2, &desc);
Chris@10: }
Chris@10: #else				/* HAVE_FMA */
Chris@10: 
Chris@10: /* Generated by: ../../../genfft/gen_twiddle.native -simd -compact -variables 4 -pipeline-latency 8 -n 2 -name t1sv_2 -include ts.h */
Chris@10: 
Chris@10: /*
Chris@10:  * This function contains 6 FP additions, 4 FP multiplications,
Chris@10:  * (or, 4 additions, 2 multiplications, 2 fused multiply/add),
Chris@10:  * 9 stack variables, 0 constants, and 8 memory accesses
Chris@10:  */
Chris@10: #include "ts.h"
Chris@10: 
Chris@10: static void t1sv_2(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@10: {
Chris@10:      {
Chris@10: 	  INT m;
Chris@10: 	  for (m = mb, W = W + (mb * 2); m < me; m = m + (2 * VL), ri = ri + ((2 * VL) * ms), ii = ii + ((2 * VL) * ms), W = W + ((2 * VL) * 2), MAKE_VOLATILE_STRIDE(4, rs)) {
Chris@10: 	       V T1, T8, T6, T7;
Chris@10: 	       T1 = LD(&(ri[0]), ms, &(ri[0]));
Chris@10: 	       T8 = LD(&(ii[0]), ms, &(ii[0]));
Chris@10: 	       {
Chris@10: 		    V T3, T5, T2, T4;
Chris@10: 		    T3 = LD(&(ri[WS(rs, 1)]), ms, &(ri[WS(rs, 1)]));
Chris@10: 		    T5 = LD(&(ii[WS(rs, 1)]), ms, &(ii[WS(rs, 1)]));
Chris@10: 		    T2 = LDW(&(W[0]));
Chris@10: 		    T4 = LDW(&(W[TWVL * 1]));
Chris@10: 		    T6 = VFMA(T2, T3, VMUL(T4, T5));
Chris@10: 		    T7 = VFNMS(T4, T3, VMUL(T2, T5));
Chris@10: 	       }
Chris@10: 	       ST(&(ri[WS(rs, 1)]), VSUB(T1, T6), ms, &(ri[WS(rs, 1)]));
Chris@10: 	       ST(&(ii[WS(rs, 1)]), VSUB(T8, T7), ms, &(ii[WS(rs, 1)]));
Chris@10: 	       ST(&(ri[0]), VADD(T1, T6), ms, &(ri[0]));
Chris@10: 	       ST(&(ii[0]), VADD(T7, T8), ms, &(ii[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:      {TW_NEXT, (2 * VL), 0}
Chris@10: };
Chris@10: 
Chris@10: static const ct_desc desc = { 2, XSIMD_STRING("t1sv_2"), twinstr, &GENUS, {4, 2, 2, 0}, 0, 0, 0 };
Chris@10: 
Chris@10: void XSIMD(codelet_t1sv_2) (planner *p) {
Chris@10:      X(kdft_dit_register) (p, t1sv_2, &desc);
Chris@10: }
Chris@10: #endif				/* HAVE_FMA */