Chris@82: /* Chris@82: * Copyright (c) 2003, 2007-14 Matteo Frigo Chris@82: * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology Chris@82: * Chris@82: * This program is free software; you can redistribute it and/or modify Chris@82: * it under the terms of the GNU General Public License as published by Chris@82: * the Free Software Foundation; either version 2 of the License, or Chris@82: * (at your option) any later version. Chris@82: * Chris@82: * This program is distributed in the hope that it will be useful, Chris@82: * but WITHOUT ANY WARRANTY; without even the implied warranty of Chris@82: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the Chris@82: * GNU General Public License for more details. Chris@82: * Chris@82: * You should have received a copy of the GNU General Public License Chris@82: * along with this program; if not, write to the Free Software Chris@82: * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA Chris@82: * Chris@82: */ Chris@82: Chris@82: /* This file was automatically generated --- DO NOT EDIT */ Chris@82: /* Generated on Thu May 24 08:06:26 EDT 2018 */ Chris@82: Chris@82: #include "rdft/codelet-rdft.h" Chris@82: Chris@82: #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA) Chris@82: Chris@82: /* Generated by: ../../../genfft/gen_r2cf.native -fma -compact -variables 4 -pipeline-latency 4 -n 7 -name r2cf_7 -include rdft/scalar/r2cf.h */ Chris@82: Chris@82: /* Chris@82: * This function contains 24 FP additions, 18 FP multiplications, Chris@82: * (or, 9 additions, 3 multiplications, 15 fused multiply/add), Chris@82: * 23 stack variables, 6 constants, and 14 memory accesses Chris@82: */ Chris@82: #include "rdft/scalar/r2cf.h" Chris@82: Chris@82: static void r2cf_7(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs) Chris@82: { Chris@82: DK(KP801937735, +0.801937735804838252472204639014890102331838324); Chris@82: DK(KP974927912, +0.974927912181823607018131682993931217232785801); Chris@82: DK(KP554958132, +0.554958132087371191422194871006410481067288862); Chris@82: DK(KP900968867, +0.900968867902419126236102319507445051165919162); Chris@82: DK(KP692021471, +0.692021471630095869627814897002069140197260599); Chris@82: DK(KP356895867, +0.356895867892209443894399510021300583399127187); Chris@82: { Chris@82: INT i; Chris@82: 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)) { Chris@82: E T1, T4, Ta, T7, Tb, Td, Tj, Ti, Th, Tf; Chris@82: T1 = R0[0]; Chris@82: { Chris@82: E T2, T3, T8, T9, T5, T6; Chris@82: T2 = R1[0]; Chris@82: T3 = R0[WS(rs, 3)]; Chris@82: T4 = T2 + T3; Chris@82: T8 = R1[WS(rs, 1)]; Chris@82: T9 = R0[WS(rs, 2)]; Chris@82: Ta = T8 + T9; Chris@82: T5 = R0[WS(rs, 1)]; Chris@82: T6 = R1[WS(rs, 2)]; Chris@82: T7 = T5 + T6; Chris@82: Tb = FNMS(KP356895867, Ta, T7); Chris@82: Td = FNMS(KP356895867, T4, Ta); Chris@82: Tj = T6 - T5; Chris@82: Ti = T9 - T8; Chris@82: Th = T3 - T2; Chris@82: Tf = FNMS(KP356895867, T7, T4); Chris@82: } Chris@82: { Chris@82: E Tc, Tm, Te, Tk, Tg, Tl; Chris@82: Tc = FNMS(KP692021471, Tb, T4); Chris@82: Cr[WS(csr, 3)] = FNMS(KP900968867, Tc, T1); Chris@82: Tm = FNMS(KP554958132, Th, Tj); Chris@82: Ci[WS(csi, 3)] = KP974927912 * (FNMS(KP801937735, Tm, Ti)); Chris@82: Te = FNMS(KP692021471, Td, T7); Chris@82: Cr[WS(csr, 2)] = FNMS(KP900968867, Te, T1); Chris@82: Tk = FMA(KP554958132, Tj, Ti); Chris@82: Ci[WS(csi, 2)] = KP974927912 * (FNMS(KP801937735, Tk, Th)); Chris@82: Cr[0] = T1 + T4 + T7 + Ta; Chris@82: Tg = FNMS(KP692021471, Tf, Ta); Chris@82: Cr[WS(csr, 1)] = FNMS(KP900968867, Tg, T1); Chris@82: Tl = FMA(KP554958132, Ti, Th); Chris@82: Ci[WS(csi, 1)] = KP974927912 * (FMA(KP801937735, Tl, Tj)); Chris@82: } Chris@82: } Chris@82: } Chris@82: } Chris@82: Chris@82: static const kr2c_desc desc = { 7, "r2cf_7", {9, 3, 15, 0}, &GENUS }; Chris@82: Chris@82: void X(codelet_r2cf_7) (planner *p) { Chris@82: X(kr2c_register) (p, r2cf_7, &desc); Chris@82: } Chris@82: Chris@82: #else Chris@82: Chris@82: /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 7 -name r2cf_7 -include rdft/scalar/r2cf.h */ Chris@82: Chris@82: /* Chris@82: * This function contains 24 FP additions, 18 FP multiplications, Chris@82: * (or, 12 additions, 6 multiplications, 12 fused multiply/add), Chris@82: * 20 stack variables, 6 constants, and 14 memory accesses Chris@82: */ Chris@82: #include "rdft/scalar/r2cf.h" Chris@82: Chris@82: static void r2cf_7(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs) Chris@82: { Chris@82: DK(KP222520933, +0.222520933956314404288902564496794759466355569); Chris@82: DK(KP900968867, +0.900968867902419126236102319507445051165919162); Chris@82: DK(KP623489801, +0.623489801858733530525004884004239810632274731); Chris@82: DK(KP433883739, +0.433883739117558120475768332848358754609990728); Chris@82: DK(KP781831482, +0.781831482468029808708444526674057750232334519); Chris@82: DK(KP974927912, +0.974927912181823607018131682993931217232785801); Chris@82: { Chris@82: INT i; Chris@82: 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)) { Chris@82: E T1, Ta, Tb, T4, Td, T7, Tc, T8, T9; Chris@82: T1 = R0[0]; Chris@82: T8 = R1[0]; Chris@82: T9 = R0[WS(rs, 3)]; Chris@82: Ta = T8 + T9; Chris@82: Tb = T9 - T8; Chris@82: { Chris@82: E T2, T3, T5, T6; Chris@82: T2 = R0[WS(rs, 1)]; Chris@82: T3 = R1[WS(rs, 2)]; Chris@82: T4 = T2 + T3; Chris@82: Td = T3 - T2; Chris@82: T5 = R1[WS(rs, 1)]; Chris@82: T6 = R0[WS(rs, 2)]; Chris@82: T7 = T5 + T6; Chris@82: Tc = T6 - T5; Chris@82: } Chris@82: Ci[WS(csi, 2)] = FNMS(KP781831482, Tc, KP974927912 * Tb) - (KP433883739 * Td); Chris@82: Ci[WS(csi, 1)] = FMA(KP781831482, Tb, KP974927912 * Td) + (KP433883739 * Tc); Chris@82: Cr[WS(csr, 2)] = FMA(KP623489801, T7, T1) + FNMA(KP900968867, T4, KP222520933 * Ta); Chris@82: Ci[WS(csi, 3)] = FMA(KP433883739, Tb, KP974927912 * Tc) - (KP781831482 * Td); Chris@82: Cr[WS(csr, 3)] = FMA(KP623489801, T4, T1) + FNMA(KP222520933, T7, KP900968867 * Ta); Chris@82: Cr[WS(csr, 1)] = FMA(KP623489801, Ta, T1) + FNMA(KP900968867, T7, KP222520933 * T4); Chris@82: Cr[0] = T1 + Ta + T4 + T7; Chris@82: } Chris@82: } Chris@82: } Chris@82: Chris@82: static const kr2c_desc desc = { 7, "r2cf_7", {12, 6, 12, 0}, &GENUS }; Chris@82: Chris@82: void X(codelet_r2cf_7) (planner *p) { Chris@82: X(kr2c_register) (p, r2cf_7, &desc); Chris@82: } Chris@82: Chris@82: #endif