cannam@127: /*
cannam@127:  * Copyright (c) 2003, 2007-14 Matteo Frigo
cannam@127:  * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
cannam@127:  *
cannam@127:  * This program is free software; you can redistribute it and/or modify
cannam@127:  * it under the terms of the GNU General Public License as published by
cannam@127:  * the Free Software Foundation; either version 2 of the License, or
cannam@127:  * (at your option) any later version.
cannam@127:  *
cannam@127:  * This program is distributed in the hope that it will be useful,
cannam@127:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
cannam@127:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
cannam@127:  * GNU General Public License for more details.
cannam@127:  *
cannam@127:  * You should have received a copy of the GNU General Public License
cannam@127:  * along with this program; if not, write to the Free Software
cannam@127:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
cannam@127:  *
cannam@127:  */
cannam@127: 
cannam@127: /* This file was automatically generated --- DO NOT EDIT */
cannam@127: /* Generated on Sat Jul 30 16:45:04 EDT 2016 */
cannam@127: 
cannam@127: #include "codelet-dft.h"
cannam@127: 
cannam@127: #ifdef HAVE_FMA
cannam@127: 
cannam@127: /* 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 */
cannam@127: 
cannam@127: /*
cannam@127:  * This function contains 6 FP additions, 4 FP multiplications,
cannam@127:  * (or, 4 additions, 2 multiplications, 2 fused multiply/add),
cannam@127:  * 11 stack variables, 0 constants, and 8 memory accesses
cannam@127:  */
cannam@127: #include "ts.h"
cannam@127: 
cannam@127: static void t1sv_2(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
cannam@127: {
cannam@127:      {
cannam@127: 	  INT m;
cannam@127: 	  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)) {
cannam@127: 	       V T1, Ta, T3, T6, T2, T5;
cannam@127: 	       T1 = LD(&(ri[0]), ms, &(ri[0]));
cannam@127: 	       Ta = LD(&(ii[0]), ms, &(ii[0]));
cannam@127: 	       T3 = LD(&(ri[WS(rs, 1)]), ms, &(ri[WS(rs, 1)]));
cannam@127: 	       T6 = LD(&(ii[WS(rs, 1)]), ms, &(ii[WS(rs, 1)]));
cannam@127: 	       T2 = LDW(&(W[0]));
cannam@127: 	       T5 = LDW(&(W[TWVL * 1]));
cannam@127: 	       {
cannam@127: 		    V T8, T4, T9, T7;
cannam@127: 		    T8 = VMUL(T2, T6);
cannam@127: 		    T4 = VMUL(T2, T3);
cannam@127: 		    T9 = VFNMS(T5, T3, T8);
cannam@127: 		    T7 = VFMA(T5, T6, T4);
cannam@127: 		    ST(&(ii[0]), VADD(T9, Ta), ms, &(ii[0]));
cannam@127: 		    ST(&(ii[WS(rs, 1)]), VSUB(Ta, T9), ms, &(ii[WS(rs, 1)]));
cannam@127: 		    ST(&(ri[0]), VADD(T1, T7), ms, &(ri[0]));
cannam@127: 		    ST(&(ri[WS(rs, 1)]), VSUB(T1, T7), ms, &(ri[WS(rs, 1)]));
cannam@127: 	       }
cannam@127: 	  }
cannam@127:      }
cannam@127:      VLEAVE();
cannam@127: }
cannam@127: 
cannam@127: static const tw_instr twinstr[] = {
cannam@127:      VTW(0, 1),
cannam@127:      {TW_NEXT, (2 * VL), 0}
cannam@127: };
cannam@127: 
cannam@127: static const ct_desc desc = { 2, XSIMD_STRING("t1sv_2"), twinstr, &GENUS, {4, 2, 2, 0}, 0, 0, 0 };
cannam@127: 
cannam@127: void XSIMD(codelet_t1sv_2) (planner *p) {
cannam@127:      X(kdft_dit_register) (p, t1sv_2, &desc);
cannam@127: }
cannam@127: #else				/* HAVE_FMA */
cannam@127: 
cannam@127: /* Generated by: ../../../genfft/gen_twiddle.native -simd -compact -variables 4 -pipeline-latency 8 -n 2 -name t1sv_2 -include ts.h */
cannam@127: 
cannam@127: /*
cannam@127:  * This function contains 6 FP additions, 4 FP multiplications,
cannam@127:  * (or, 4 additions, 2 multiplications, 2 fused multiply/add),
cannam@127:  * 9 stack variables, 0 constants, and 8 memory accesses
cannam@127:  */
cannam@127: #include "ts.h"
cannam@127: 
cannam@127: static void t1sv_2(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
cannam@127: {
cannam@127:      {
cannam@127: 	  INT m;
cannam@127: 	  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)) {
cannam@127: 	       V T1, T8, T6, T7;
cannam@127: 	       T1 = LD(&(ri[0]), ms, &(ri[0]));
cannam@127: 	       T8 = LD(&(ii[0]), ms, &(ii[0]));
cannam@127: 	       {
cannam@127: 		    V T3, T5, T2, T4;
cannam@127: 		    T3 = LD(&(ri[WS(rs, 1)]), ms, &(ri[WS(rs, 1)]));
cannam@127: 		    T5 = LD(&(ii[WS(rs, 1)]), ms, &(ii[WS(rs, 1)]));
cannam@127: 		    T2 = LDW(&(W[0]));
cannam@127: 		    T4 = LDW(&(W[TWVL * 1]));
cannam@127: 		    T6 = VFMA(T2, T3, VMUL(T4, T5));
cannam@127: 		    T7 = VFNMS(T4, T3, VMUL(T2, T5));
cannam@127: 	       }
cannam@127: 	       ST(&(ri[WS(rs, 1)]), VSUB(T1, T6), ms, &(ri[WS(rs, 1)]));
cannam@127: 	       ST(&(ii[WS(rs, 1)]), VSUB(T8, T7), ms, &(ii[WS(rs, 1)]));
cannam@127: 	       ST(&(ri[0]), VADD(T1, T6), ms, &(ri[0]));
cannam@127: 	       ST(&(ii[0]), VADD(T7, T8), ms, &(ii[0]));
cannam@127: 	  }
cannam@127:      }
cannam@127:      VLEAVE();
cannam@127: }
cannam@127: 
cannam@127: static const tw_instr twinstr[] = {
cannam@127:      VTW(0, 1),
cannam@127:      {TW_NEXT, (2 * VL), 0}
cannam@127: };
cannam@127: 
cannam@127: static const ct_desc desc = { 2, XSIMD_STRING("t1sv_2"), twinstr, &GENUS, {4, 2, 2, 0}, 0, 0, 0 };
cannam@127: 
cannam@127: void XSIMD(codelet_t1sv_2) (planner *p) {
cannam@127:      X(kdft_dit_register) (p, t1sv_2, &desc);
cannam@127: }
cannam@127: #endif				/* HAVE_FMA */