cannam@167: /* cannam@167: * Copyright (c) 2003, 2007-14 Matteo Frigo cannam@167: * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology cannam@167: * cannam@167: * This program is free software; you can redistribute it and/or modify cannam@167: * it under the terms of the GNU General Public License as published by cannam@167: * the Free Software Foundation; either version 2 of the License, or cannam@167: * (at your option) any later version. cannam@167: * cannam@167: * This program is distributed in the hope that it will be useful, cannam@167: * but WITHOUT ANY WARRANTY; without even the implied warranty of cannam@167: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the cannam@167: * GNU General Public License for more details. cannam@167: * cannam@167: * You should have received a copy of the GNU General Public License cannam@167: * along with this program; if not, write to the Free Software cannam@167: * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA cannam@167: * cannam@167: */ cannam@167: cannam@167: /* This file was automatically generated --- DO NOT EDIT */ cannam@167: /* Generated on Thu May 24 08:07:42 EDT 2018 */ cannam@167: cannam@167: #include "rdft/codelet-rdft.h" cannam@167: cannam@167: #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA) cannam@167: cannam@167: /* Generated by: ../../../genfft/gen_r2cb.native -fma -compact -variables 4 -pipeline-latency 4 -sign 1 -n 3 -name r2cbIII_3 -dft-III -include rdft/scalar/r2cbIII.h */ cannam@167: cannam@167: /* cannam@167: * This function contains 4 FP additions, 3 FP multiplications, cannam@167: * (or, 1 additions, 0 multiplications, 3 fused multiply/add), cannam@167: * 7 stack variables, 2 constants, and 6 memory accesses cannam@167: */ cannam@167: #include "rdft/scalar/r2cbIII.h" cannam@167: cannam@167: static void r2cbIII_3(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs) cannam@167: { cannam@167: DK(KP1_732050807, +1.732050807568877293527446341505872366942805254); cannam@167: DK(KP2_000000000, +2.000000000000000000000000000000000000000000000); cannam@167: { cannam@167: INT i; cannam@167: for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(12, rs), MAKE_VOLATILE_STRIDE(12, csr), MAKE_VOLATILE_STRIDE(12, csi)) { cannam@167: E T4, T1, T2, T3; cannam@167: T4 = Ci[0]; cannam@167: T1 = Cr[WS(csr, 1)]; cannam@167: T2 = Cr[0]; cannam@167: T3 = T2 - T1; cannam@167: R0[0] = FMA(KP2_000000000, T2, T1); cannam@167: R0[WS(rs, 1)] = -(FMA(KP1_732050807, T4, T3)); cannam@167: R1[0] = FNMS(KP1_732050807, T4, T3); cannam@167: } cannam@167: } cannam@167: } cannam@167: cannam@167: static const kr2c_desc desc = { 3, "r2cbIII_3", {1, 0, 3, 0}, &GENUS }; cannam@167: cannam@167: void X(codelet_r2cbIII_3) (planner *p) { cannam@167: X(kr2c_register) (p, r2cbIII_3, &desc); cannam@167: } cannam@167: cannam@167: #else cannam@167: cannam@167: /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 3 -name r2cbIII_3 -dft-III -include rdft/scalar/r2cbIII.h */ cannam@167: cannam@167: /* cannam@167: * This function contains 4 FP additions, 2 FP multiplications, cannam@167: * (or, 3 additions, 1 multiplications, 1 fused multiply/add), cannam@167: * 8 stack variables, 2 constants, and 6 memory accesses cannam@167: */ cannam@167: #include "rdft/scalar/r2cbIII.h" cannam@167: cannam@167: static void r2cbIII_3(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs) cannam@167: { cannam@167: DK(KP2_000000000, +2.000000000000000000000000000000000000000000000); cannam@167: DK(KP1_732050807, +1.732050807568877293527446341505872366942805254); cannam@167: { cannam@167: INT i; cannam@167: for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(12, rs), MAKE_VOLATILE_STRIDE(12, csr), MAKE_VOLATILE_STRIDE(12, csi)) { cannam@167: E T5, T1, T2, T3, T4; cannam@167: T4 = Ci[0]; cannam@167: T5 = KP1_732050807 * T4; cannam@167: T1 = Cr[WS(csr, 1)]; cannam@167: T2 = Cr[0]; cannam@167: T3 = T2 - T1; cannam@167: R0[0] = FMA(KP2_000000000, T2, T1); cannam@167: R0[WS(rs, 1)] = -(T3 + T5); cannam@167: R1[0] = T3 - T5; cannam@167: } cannam@167: } cannam@167: } cannam@167: cannam@167: static const kr2c_desc desc = { 3, "r2cbIII_3", {3, 1, 1, 0}, &GENUS }; cannam@167: cannam@167: void X(codelet_r2cbIII_3) (planner *p) { cannam@167: X(kr2c_register) (p, r2cbIII_3, &desc); cannam@167: } cannam@167: cannam@167: #endif