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 07:39:45 EST 2012 */ cannam@95: cannam@95: #include "codelet-rdft.h" cannam@95: cannam@95: #ifdef HAVE_FMA cannam@95: cannam@95: /* Generated by: ../../../genfft/gen_r2cf.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 7 -name r2cf_7 -include r2cf.h */ cannam@95: cannam@95: /* cannam@95: * This function contains 24 FP additions, 18 FP multiplications, cannam@95: * (or, 9 additions, 3 multiplications, 15 fused multiply/add), cannam@95: * 25 stack variables, 6 constants, and 14 memory accesses cannam@95: */ cannam@95: #include "r2cf.h" cannam@95: cannam@95: static void r2cf_7(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs) cannam@95: { cannam@95: DK(KP900968867, +0.900968867902419126236102319507445051165919162); cannam@95: DK(KP801937735, +0.801937735804838252472204639014890102331838324); cannam@95: DK(KP974927912, +0.974927912181823607018131682993931217232785801); cannam@95: DK(KP692021471, +0.692021471630095869627814897002069140197260599); cannam@95: DK(KP554958132, +0.554958132087371191422194871006410481067288862); cannam@95: DK(KP356895867, +0.356895867892209443894399510021300583399127187); cannam@95: { cannam@95: INT i; cannam@95: for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(28, rs), MAKE_VOLATILE_STRIDE(28, csr), MAKE_VOLATILE_STRIDE(28, csi)) { cannam@95: E T1, Tg, Tc; cannam@95: { cannam@95: E Th, T4, Ti, Ta, Tj, T7, Td, T5, T6, Tl, Tk; cannam@95: T1 = R0[0]; cannam@95: { cannam@95: E T2, T3, T8, T9; cannam@95: T2 = R1[0]; cannam@95: T3 = R0[WS(rs, 3)]; cannam@95: T8 = R1[WS(rs, 1)]; cannam@95: T9 = R0[WS(rs, 2)]; cannam@95: T5 = R0[WS(rs, 1)]; cannam@95: Th = T3 - T2; cannam@95: T4 = T2 + T3; cannam@95: T6 = R1[WS(rs, 2)]; cannam@95: Ti = T9 - T8; cannam@95: Ta = T8 + T9; cannam@95: } cannam@95: Tj = T6 - T5; cannam@95: T7 = T5 + T6; cannam@95: Td = FNMS(KP356895867, T4, Ta); cannam@95: Tl = FMA(KP554958132, Ti, Th); cannam@95: Tk = FMA(KP554958132, Tj, Ti); cannam@95: { cannam@95: E Tm, Tf, Tb, Te; cannam@95: Tm = FNMS(KP554958132, Th, Tj); cannam@95: Cr[0] = T1 + T4 + T7 + Ta; cannam@95: Tf = FNMS(KP356895867, T7, T4); cannam@95: Tb = FNMS(KP356895867, Ta, T7); cannam@95: Te = FNMS(KP692021471, Td, T7); cannam@95: Ci[WS(csi, 2)] = KP974927912 * (FNMS(KP801937735, Tk, Th)); cannam@95: Ci[WS(csi, 3)] = KP974927912 * (FNMS(KP801937735, Tm, Ti)); cannam@95: Tg = FNMS(KP692021471, Tf, Ta); cannam@95: Tc = FNMS(KP692021471, Tb, T4); cannam@95: Cr[WS(csr, 2)] = FNMS(KP900968867, Te, T1); cannam@95: Ci[WS(csi, 1)] = KP974927912 * (FMA(KP801937735, Tl, Tj)); cannam@95: } cannam@95: } cannam@95: Cr[WS(csr, 1)] = FNMS(KP900968867, Tg, T1); cannam@95: Cr[WS(csr, 3)] = FNMS(KP900968867, Tc, T1); cannam@95: } cannam@95: } cannam@95: } cannam@95: cannam@95: static const kr2c_desc desc = { 7, "r2cf_7", {9, 3, 15, 0}, &GENUS }; cannam@95: cannam@95: void X(codelet_r2cf_7) (planner *p) { cannam@95: X(kr2c_register) (p, r2cf_7, &desc); cannam@95: } cannam@95: cannam@95: #else /* HAVE_FMA */ cannam@95: cannam@95: /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 7 -name r2cf_7 -include r2cf.h */ cannam@95: cannam@95: /* cannam@95: * This function contains 24 FP additions, 18 FP multiplications, cannam@95: * (or, 12 additions, 6 multiplications, 12 fused multiply/add), cannam@95: * 20 stack variables, 6 constants, and 14 memory accesses cannam@95: */ cannam@95: #include "r2cf.h" cannam@95: cannam@95: static void r2cf_7(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs) cannam@95: { cannam@95: DK(KP222520933, +0.222520933956314404288902564496794759466355569); cannam@95: DK(KP900968867, +0.900968867902419126236102319507445051165919162); cannam@95: DK(KP623489801, +0.623489801858733530525004884004239810632274731); cannam@95: DK(KP433883739, +0.433883739117558120475768332848358754609990728); cannam@95: DK(KP781831482, +0.781831482468029808708444526674057750232334519); cannam@95: DK(KP974927912, +0.974927912181823607018131682993931217232785801); cannam@95: { cannam@95: INT i; cannam@95: for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(28, rs), MAKE_VOLATILE_STRIDE(28, csr), MAKE_VOLATILE_STRIDE(28, csi)) { cannam@95: E T1, Ta, Tb, T4, Td, T7, Tc, T8, T9; cannam@95: T1 = R0[0]; cannam@95: T8 = R1[0]; cannam@95: T9 = R0[WS(rs, 3)]; cannam@95: Ta = T8 + T9; cannam@95: Tb = T9 - T8; cannam@95: { cannam@95: E T2, T3, T5, T6; cannam@95: T2 = R0[WS(rs, 1)]; cannam@95: T3 = R1[WS(rs, 2)]; cannam@95: T4 = T2 + T3; cannam@95: Td = T3 - T2; cannam@95: T5 = R1[WS(rs, 1)]; cannam@95: T6 = R0[WS(rs, 2)]; cannam@95: T7 = T5 + T6; cannam@95: Tc = T6 - T5; cannam@95: } cannam@95: Ci[WS(csi, 2)] = FNMS(KP781831482, Tc, KP974927912 * Tb) - (KP433883739 * Td); cannam@95: Ci[WS(csi, 1)] = FMA(KP781831482, Tb, KP974927912 * Td) + (KP433883739 * Tc); cannam@95: Cr[WS(csr, 2)] = FMA(KP623489801, T7, T1) + FNMA(KP900968867, T4, KP222520933 * Ta); cannam@95: Ci[WS(csi, 3)] = FMA(KP433883739, Tb, KP974927912 * Tc) - (KP781831482 * Td); cannam@95: Cr[WS(csr, 3)] = FMA(KP623489801, T4, T1) + FNMA(KP222520933, T7, KP900968867 * Ta); cannam@95: Cr[WS(csr, 1)] = FMA(KP623489801, Ta, T1) + FNMA(KP900968867, T7, KP222520933 * T4); cannam@95: Cr[0] = T1 + Ta + T4 + T7; cannam@95: } cannam@95: } cannam@95: } cannam@95: cannam@95: static const kr2c_desc desc = { 7, "r2cf_7", {12, 6, 12, 0}, &GENUS }; cannam@95: cannam@95: void X(codelet_r2cf_7) (planner *p) { cannam@95: X(kr2c_register) (p, r2cf_7, &desc); cannam@95: } cannam@95: cannam@95: #endif /* HAVE_FMA */