cannam@95: /* cannam@95: * Copyright (c) 2003, 2007-11 Matteo Frigo cannam@95: * Copyright (c) 2003, 2007-11 Massachusetts Institute of Technology cannam@95: * cannam@95: * This program is free software; you can redistribute it and/or modify cannam@95: * it under the terms of the GNU General Public License as published by cannam@95: * the Free Software Foundation; either version 2 of the License, or cannam@95: * (at your option) any later version. cannam@95: * cannam@95: * This program is distributed in the hope that it will be useful, cannam@95: * but WITHOUT ANY WARRANTY; without even the implied warranty of cannam@95: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the cannam@95: * GNU General Public License for more details. cannam@95: * cannam@95: * You should have received a copy of the GNU General Public License cannam@95: * along with this program; if not, write to the Free Software cannam@95: * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA cannam@95: * cannam@95: */ cannam@95: cannam@95: /* This file was automatically generated --- DO NOT EDIT */ cannam@95: /* Generated on Sun Nov 25 07:39:23 EST 2012 */ cannam@95: cannam@95: #include "codelet-dft.h" cannam@95: cannam@95: #ifdef HAVE_FMA cannam@95: cannam@95: /* 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@95: cannam@95: /* cannam@95: * This function contains 6 FP additions, 4 FP multiplications, cannam@95: * (or, 4 additions, 2 multiplications, 2 fused multiply/add), cannam@95: * 11 stack variables, 0 constants, and 8 memory accesses cannam@95: */ cannam@95: #include "ts.h" cannam@95: cannam@95: static void t1sv_2(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) cannam@95: { cannam@95: { cannam@95: INT m; cannam@95: 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@95: V T1, Ta, T3, T6, T2, T5; cannam@95: T1 = LD(&(ri[0]), ms, &(ri[0])); cannam@95: Ta = LD(&(ii[0]), ms, &(ii[0])); cannam@95: T3 = LD(&(ri[WS(rs, 1)]), ms, &(ri[WS(rs, 1)])); cannam@95: T6 = LD(&(ii[WS(rs, 1)]), ms, &(ii[WS(rs, 1)])); cannam@95: T2 = LDW(&(W[0])); cannam@95: T5 = LDW(&(W[TWVL * 1])); cannam@95: { cannam@95: V T8, T4, T9, T7; cannam@95: T8 = VMUL(T2, T6); cannam@95: T4 = VMUL(T2, T3); cannam@95: T9 = VFNMS(T5, T3, T8); cannam@95: T7 = VFMA(T5, T6, T4); cannam@95: ST(&(ii[0]), VADD(T9, Ta), ms, &(ii[0])); cannam@95: ST(&(ii[WS(rs, 1)]), VSUB(Ta, T9), ms, &(ii[WS(rs, 1)])); cannam@95: ST(&(ri[0]), VADD(T1, T7), ms, &(ri[0])); cannam@95: ST(&(ri[WS(rs, 1)]), VSUB(T1, T7), ms, &(ri[WS(rs, 1)])); cannam@95: } cannam@95: } cannam@95: } cannam@95: VLEAVE(); cannam@95: } cannam@95: cannam@95: static const tw_instr twinstr[] = { cannam@95: VTW(0, 1), cannam@95: {TW_NEXT, (2 * VL), 0} cannam@95: }; cannam@95: cannam@95: static const ct_desc desc = { 2, XSIMD_STRING("t1sv_2"), twinstr, &GENUS, {4, 2, 2, 0}, 0, 0, 0 }; cannam@95: cannam@95: void XSIMD(codelet_t1sv_2) (planner *p) { cannam@95: X(kdft_dit_register) (p, t1sv_2, &desc); cannam@95: } cannam@95: #else /* HAVE_FMA */ cannam@95: cannam@95: /* Generated by: ../../../genfft/gen_twiddle.native -simd -compact -variables 4 -pipeline-latency 8 -n 2 -name t1sv_2 -include ts.h */ cannam@95: cannam@95: /* cannam@95: * This function contains 6 FP additions, 4 FP multiplications, cannam@95: * (or, 4 additions, 2 multiplications, 2 fused multiply/add), cannam@95: * 9 stack variables, 0 constants, and 8 memory accesses cannam@95: */ cannam@95: #include "ts.h" cannam@95: cannam@95: static void t1sv_2(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) cannam@95: { cannam@95: { cannam@95: INT m; cannam@95: 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@95: V T1, T8, T6, T7; cannam@95: T1 = LD(&(ri[0]), ms, &(ri[0])); cannam@95: T8 = LD(&(ii[0]), ms, &(ii[0])); cannam@95: { cannam@95: V T3, T5, T2, T4; cannam@95: T3 = LD(&(ri[WS(rs, 1)]), ms, &(ri[WS(rs, 1)])); cannam@95: T5 = LD(&(ii[WS(rs, 1)]), ms, &(ii[WS(rs, 1)])); cannam@95: T2 = LDW(&(W[0])); cannam@95: T4 = LDW(&(W[TWVL * 1])); cannam@95: T6 = VFMA(T2, T3, VMUL(T4, T5)); cannam@95: T7 = VFNMS(T4, T3, VMUL(T2, T5)); cannam@95: } cannam@95: ST(&(ri[WS(rs, 1)]), VSUB(T1, T6), ms, &(ri[WS(rs, 1)])); cannam@95: ST(&(ii[WS(rs, 1)]), VSUB(T8, T7), ms, &(ii[WS(rs, 1)])); cannam@95: ST(&(ri[0]), VADD(T1, T6), ms, &(ri[0])); cannam@95: ST(&(ii[0]), VADD(T7, T8), ms, &(ii[0])); cannam@95: } cannam@95: } cannam@95: VLEAVE(); cannam@95: } cannam@95: cannam@95: static const tw_instr twinstr[] = { cannam@95: VTW(0, 1), cannam@95: {TW_NEXT, (2 * VL), 0} cannam@95: }; cannam@95: cannam@95: static const ct_desc desc = { 2, XSIMD_STRING("t1sv_2"), twinstr, &GENUS, {4, 2, 2, 0}, 0, 0, 0 }; cannam@95: cannam@95: void XSIMD(codelet_t1sv_2) (planner *p) { cannam@95: X(kdft_dit_register) (p, t1sv_2, &desc); cannam@95: } cannam@95: #endif /* HAVE_FMA */