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 08:04:10 EDT 2018 */ cannam@167: cannam@167: #include "dft/codelet-dft.h" cannam@167: cannam@167: #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA) cannam@167: cannam@167: /* Generated by: ../../../genfft/gen_notw.native -fma -compact -variables 4 -pipeline-latency 4 -n 8 -name n1_8 -include dft/scalar/n.h */ cannam@167: cannam@167: /* cannam@167: * This function contains 52 FP additions, 8 FP multiplications, cannam@167: * (or, 44 additions, 0 multiplications, 8 fused multiply/add), cannam@167: * 28 stack variables, 1 constants, and 32 memory accesses cannam@167: */ cannam@167: #include "dft/scalar/n.h" cannam@167: cannam@167: static void n1_8(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) cannam@167: { cannam@167: DK(KP707106781, +0.707106781186547524400844362104849039284835938); cannam@167: { cannam@167: INT i; cannam@167: for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(32, is), MAKE_VOLATILE_STRIDE(32, os)) { cannam@167: E T3, Tn, Ti, TC, T6, TB, Tl, To, Td, TN, Tz, TH, Ta, TM, Tu; cannam@167: E TG; cannam@167: { cannam@167: E T1, T2, Tj, Tk; cannam@167: T1 = ri[0]; cannam@167: T2 = ri[WS(is, 4)]; cannam@167: T3 = T1 + T2; cannam@167: Tn = T1 - T2; cannam@167: { cannam@167: E Tg, Th, T4, T5; cannam@167: Tg = ii[0]; cannam@167: Th = ii[WS(is, 4)]; cannam@167: Ti = Tg + Th; cannam@167: TC = Tg - Th; cannam@167: T4 = ri[WS(is, 2)]; cannam@167: T5 = ri[WS(is, 6)]; cannam@167: T6 = T4 + T5; cannam@167: TB = T4 - T5; cannam@167: } cannam@167: Tj = ii[WS(is, 2)]; cannam@167: Tk = ii[WS(is, 6)]; cannam@167: Tl = Tj + Tk; cannam@167: To = Tj - Tk; cannam@167: { cannam@167: E Tb, Tc, Tv, Tw, Tx, Ty; cannam@167: Tb = ri[WS(is, 7)]; cannam@167: Tc = ri[WS(is, 3)]; cannam@167: Tv = Tb - Tc; cannam@167: Tw = ii[WS(is, 7)]; cannam@167: Tx = ii[WS(is, 3)]; cannam@167: Ty = Tw - Tx; cannam@167: Td = Tb + Tc; cannam@167: TN = Tw + Tx; cannam@167: Tz = Tv - Ty; cannam@167: TH = Tv + Ty; cannam@167: } cannam@167: { cannam@167: E T8, T9, Tq, Tr, Ts, Tt; cannam@167: T8 = ri[WS(is, 1)]; cannam@167: T9 = ri[WS(is, 5)]; cannam@167: Tq = T8 - T9; cannam@167: Tr = ii[WS(is, 1)]; cannam@167: Ts = ii[WS(is, 5)]; cannam@167: Tt = Tr - Ts; cannam@167: Ta = T8 + T9; cannam@167: TM = Tr + Ts; cannam@167: Tu = Tq + Tt; cannam@167: TG = Tt - Tq; cannam@167: } cannam@167: } cannam@167: { cannam@167: E T7, Te, TP, TQ; cannam@167: T7 = T3 + T6; cannam@167: Te = Ta + Td; cannam@167: ro[WS(os, 4)] = T7 - Te; cannam@167: ro[0] = T7 + Te; cannam@167: TP = Ti + Tl; cannam@167: TQ = TM + TN; cannam@167: io[WS(os, 4)] = TP - TQ; cannam@167: io[0] = TP + TQ; cannam@167: } cannam@167: { cannam@167: E Tf, Tm, TL, TO; cannam@167: Tf = Td - Ta; cannam@167: Tm = Ti - Tl; cannam@167: io[WS(os, 2)] = Tf + Tm; cannam@167: io[WS(os, 6)] = Tm - Tf; cannam@167: TL = T3 - T6; cannam@167: TO = TM - TN; cannam@167: ro[WS(os, 6)] = TL - TO; cannam@167: ro[WS(os, 2)] = TL + TO; cannam@167: } cannam@167: { cannam@167: E Tp, TA, TJ, TK; cannam@167: Tp = Tn + To; cannam@167: TA = Tu + Tz; cannam@167: ro[WS(os, 5)] = FNMS(KP707106781, TA, Tp); cannam@167: ro[WS(os, 1)] = FMA(KP707106781, TA, Tp); cannam@167: TJ = TC - TB; cannam@167: TK = TG + TH; cannam@167: io[WS(os, 5)] = FNMS(KP707106781, TK, TJ); cannam@167: io[WS(os, 1)] = FMA(KP707106781, TK, TJ); cannam@167: } cannam@167: { cannam@167: E TD, TE, TF, TI; cannam@167: TD = TB + TC; cannam@167: TE = Tz - Tu; cannam@167: io[WS(os, 7)] = FNMS(KP707106781, TE, TD); cannam@167: io[WS(os, 3)] = FMA(KP707106781, TE, TD); cannam@167: TF = Tn - To; cannam@167: TI = TG - TH; cannam@167: ro[WS(os, 7)] = FNMS(KP707106781, TI, TF); cannam@167: ro[WS(os, 3)] = FMA(KP707106781, TI, TF); cannam@167: } cannam@167: } cannam@167: } cannam@167: } cannam@167: cannam@167: static const kdft_desc desc = { 8, "n1_8", {44, 0, 8, 0}, &GENUS, 0, 0, 0, 0 }; cannam@167: cannam@167: void X(codelet_n1_8) (planner *p) { cannam@167: X(kdft_register) (p, n1_8, &desc); cannam@167: } cannam@167: cannam@167: #else cannam@167: cannam@167: /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 8 -name n1_8 -include dft/scalar/n.h */ cannam@167: cannam@167: /* cannam@167: * This function contains 52 FP additions, 4 FP multiplications, cannam@167: * (or, 52 additions, 4 multiplications, 0 fused multiply/add), cannam@167: * 28 stack variables, 1 constants, and 32 memory accesses cannam@167: */ cannam@167: #include "dft/scalar/n.h" cannam@167: cannam@167: static void n1_8(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) cannam@167: { cannam@167: DK(KP707106781, +0.707106781186547524400844362104849039284835938); cannam@167: { cannam@167: INT i; cannam@167: for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(32, is), MAKE_VOLATILE_STRIDE(32, os)) { cannam@167: E T3, Tn, Ti, TC, T6, TB, Tl, To, Td, TN, Tz, TH, Ta, TM, Tu; cannam@167: E TG; cannam@167: { cannam@167: E T1, T2, Tj, Tk; cannam@167: T1 = ri[0]; cannam@167: T2 = ri[WS(is, 4)]; cannam@167: T3 = T1 + T2; cannam@167: Tn = T1 - T2; cannam@167: { cannam@167: E Tg, Th, T4, T5; cannam@167: Tg = ii[0]; cannam@167: Th = ii[WS(is, 4)]; cannam@167: Ti = Tg + Th; cannam@167: TC = Tg - Th; cannam@167: T4 = ri[WS(is, 2)]; cannam@167: T5 = ri[WS(is, 6)]; cannam@167: T6 = T4 + T5; cannam@167: TB = T4 - T5; cannam@167: } cannam@167: Tj = ii[WS(is, 2)]; cannam@167: Tk = ii[WS(is, 6)]; cannam@167: Tl = Tj + Tk; cannam@167: To = Tj - Tk; cannam@167: { cannam@167: E Tb, Tc, Tv, Tw, Tx, Ty; cannam@167: Tb = ri[WS(is, 7)]; cannam@167: Tc = ri[WS(is, 3)]; cannam@167: Tv = Tb - Tc; cannam@167: Tw = ii[WS(is, 7)]; cannam@167: Tx = ii[WS(is, 3)]; cannam@167: Ty = Tw - Tx; cannam@167: Td = Tb + Tc; cannam@167: TN = Tw + Tx; cannam@167: Tz = Tv - Ty; cannam@167: TH = Tv + Ty; cannam@167: } cannam@167: { cannam@167: E T8, T9, Tq, Tr, Ts, Tt; cannam@167: T8 = ri[WS(is, 1)]; cannam@167: T9 = ri[WS(is, 5)]; cannam@167: Tq = T8 - T9; cannam@167: Tr = ii[WS(is, 1)]; cannam@167: Ts = ii[WS(is, 5)]; cannam@167: Tt = Tr - Ts; cannam@167: Ta = T8 + T9; cannam@167: TM = Tr + Ts; cannam@167: Tu = Tq + Tt; cannam@167: TG = Tt - Tq; cannam@167: } cannam@167: } cannam@167: { cannam@167: E T7, Te, TP, TQ; cannam@167: T7 = T3 + T6; cannam@167: Te = Ta + Td; cannam@167: ro[WS(os, 4)] = T7 - Te; cannam@167: ro[0] = T7 + Te; cannam@167: TP = Ti + Tl; cannam@167: TQ = TM + TN; cannam@167: io[WS(os, 4)] = TP - TQ; cannam@167: io[0] = TP + TQ; cannam@167: } cannam@167: { cannam@167: E Tf, Tm, TL, TO; cannam@167: Tf = Td - Ta; cannam@167: Tm = Ti - Tl; cannam@167: io[WS(os, 2)] = Tf + Tm; cannam@167: io[WS(os, 6)] = Tm - Tf; cannam@167: TL = T3 - T6; cannam@167: TO = TM - TN; cannam@167: ro[WS(os, 6)] = TL - TO; cannam@167: ro[WS(os, 2)] = TL + TO; cannam@167: } cannam@167: { cannam@167: E Tp, TA, TJ, TK; cannam@167: Tp = Tn + To; cannam@167: TA = KP707106781 * (Tu + Tz); cannam@167: ro[WS(os, 5)] = Tp - TA; cannam@167: ro[WS(os, 1)] = Tp + TA; cannam@167: TJ = TC - TB; cannam@167: TK = KP707106781 * (TG + TH); cannam@167: io[WS(os, 5)] = TJ - TK; cannam@167: io[WS(os, 1)] = TJ + TK; cannam@167: } cannam@167: { cannam@167: E TD, TE, TF, TI; cannam@167: TD = TB + TC; cannam@167: TE = KP707106781 * (Tz - Tu); cannam@167: io[WS(os, 7)] = TD - TE; cannam@167: io[WS(os, 3)] = TD + TE; cannam@167: TF = Tn - To; cannam@167: TI = KP707106781 * (TG - TH); cannam@167: ro[WS(os, 7)] = TF - TI; cannam@167: ro[WS(os, 3)] = TF + TI; cannam@167: } cannam@167: } cannam@167: } cannam@167: } cannam@167: cannam@167: static const kdft_desc desc = { 8, "n1_8", {52, 4, 0, 0}, &GENUS, 0, 0, 0, 0 }; cannam@167: cannam@167: void X(codelet_n1_8) (planner *p) { cannam@167: X(kdft_register) (p, n1_8, &desc); cannam@167: } cannam@167: cannam@167: #endif