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 08:06:26 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_r2cf.native -fma -compact -variables 4 -pipeline-latency 4 -n 15 -name r2cf_15 -include rdft/scalar/r2cf.h */ cannam@167: cannam@167: /* cannam@167: * This function contains 64 FP additions, 35 FP multiplications, cannam@167: * (or, 36 additions, 7 multiplications, 28 fused multiply/add), cannam@167: * 45 stack variables, 8 constants, and 30 memory accesses cannam@167: */ cannam@167: #include "rdft/scalar/r2cf.h" cannam@167: cannam@167: static void r2cf_15(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(KP910592997, +0.910592997310029334643087372129977886038870291); cannam@167: DK(KP951056516, +0.951056516295153572116439333379382143405698634); cannam@167: DK(KP823639103, +0.823639103546331925877420039278190003029660514); cannam@167: DK(KP559016994, +0.559016994374947424102293417182819058860154590); cannam@167: DK(KP250000000, +0.250000000000000000000000000000000000000000000); cannam@167: DK(KP618033988, +0.618033988749894848204586834365638117720309180); cannam@167: DK(KP866025403, +0.866025403784438646763723170752936183471402627); cannam@167: DK(KP500000000, +0.500000000000000000000000000000000000000000000); cannam@167: { cannam@167: INT i; cannam@167: for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(60, rs), MAKE_VOLATILE_STRIDE(60, csr), MAKE_VOLATILE_STRIDE(60, csi)) { cannam@167: E Ti, TR, TF, TM, TN, T7, Te, Tf, TV, TW, TX, Ts, Tv, TH, Tl; cannam@167: E To, TG, TS, TT, TU; cannam@167: { cannam@167: E TD, Tg, Th, TE; cannam@167: TD = R0[0]; cannam@167: Tg = R0[WS(rs, 5)]; cannam@167: Th = R1[WS(rs, 2)]; cannam@167: TE = Th + Tg; cannam@167: Ti = Tg - Th; cannam@167: TR = TD + TE; cannam@167: TF = FNMS(KP500000000, TE, TD); cannam@167: } cannam@167: { cannam@167: E Tj, Tq, Tt, Tm, T3, Tk, Ta, Tr, Td, Tu, T6, Tn; cannam@167: Tj = R1[WS(rs, 1)]; cannam@167: Tq = R0[WS(rs, 3)]; cannam@167: Tt = R1[WS(rs, 4)]; cannam@167: Tm = R0[WS(rs, 6)]; cannam@167: { cannam@167: E T1, T2, T8, T9; cannam@167: T1 = R0[WS(rs, 4)]; cannam@167: T2 = R1[WS(rs, 6)]; cannam@167: T3 = T1 - T2; cannam@167: Tk = T1 + T2; cannam@167: T8 = R1[WS(rs, 5)]; cannam@167: T9 = R1[0]; cannam@167: Ta = T8 - T9; cannam@167: Tr = T8 + T9; cannam@167: } cannam@167: { cannam@167: E Tb, Tc, T4, T5; cannam@167: Tb = R0[WS(rs, 7)]; cannam@167: Tc = R0[WS(rs, 2)]; cannam@167: Td = Tb - Tc; cannam@167: Tu = Tb + Tc; cannam@167: T4 = R0[WS(rs, 1)]; cannam@167: T5 = R1[WS(rs, 3)]; cannam@167: T6 = T4 - T5; cannam@167: Tn = T4 + T5; cannam@167: } cannam@167: TM = T6 - T3; cannam@167: TN = Td - Ta; cannam@167: T7 = T3 + T6; cannam@167: Te = Ta + Td; cannam@167: Tf = T7 + Te; cannam@167: TV = Tq + Tr; cannam@167: TW = Tt + Tu; cannam@167: TX = TV + TW; cannam@167: Ts = FNMS(KP500000000, Tr, Tq); cannam@167: Tv = FNMS(KP500000000, Tu, Tt); cannam@167: TH = Ts + Tv; cannam@167: Tl = FNMS(KP500000000, Tk, Tj); cannam@167: To = FNMS(KP500000000, Tn, Tm); cannam@167: TG = Tl + To; cannam@167: TS = Tj + Tk; cannam@167: TT = Tm + Tn; cannam@167: TU = TS + TT; cannam@167: } cannam@167: Ci[WS(csi, 5)] = KP866025403 * (Tf - Ti); cannam@167: { cannam@167: E TK, TQ, TO, TI, TJ, TP, TL; cannam@167: TK = TG - TH; cannam@167: TQ = FNMS(KP618033988, TM, TN); cannam@167: TO = FMA(KP618033988, TN, TM); cannam@167: TI = TG + TH; cannam@167: TJ = FNMS(KP250000000, TI, TF); cannam@167: Cr[WS(csr, 5)] = TF + TI; cannam@167: TP = FNMS(KP559016994, TK, TJ); cannam@167: Cr[WS(csr, 2)] = FMA(KP823639103, TQ, TP); cannam@167: Cr[WS(csr, 7)] = FNMS(KP823639103, TQ, TP); cannam@167: TL = FMA(KP559016994, TK, TJ); cannam@167: Cr[WS(csr, 1)] = FMA(KP823639103, TO, TL); cannam@167: Cr[WS(csr, 4)] = FNMS(KP823639103, TO, TL); cannam@167: } cannam@167: { cannam@167: E T11, T12, T10, TY, TZ; cannam@167: T11 = TW - TV; cannam@167: T12 = TS - TT; cannam@167: Ci[WS(csi, 3)] = KP951056516 * (FMA(KP618033988, T12, T11)); cannam@167: Ci[WS(csi, 6)] = -(KP951056516 * (FNMS(KP618033988, T11, T12))); cannam@167: T10 = TU - TX; cannam@167: TY = TU + TX; cannam@167: TZ = FNMS(KP250000000, TY, TR); cannam@167: Cr[WS(csr, 3)] = FNMS(KP559016994, T10, TZ); cannam@167: Cr[0] = TR + TY; cannam@167: Cr[WS(csr, 6)] = FMA(KP559016994, T10, TZ); cannam@167: { cannam@167: E Tx, TB, TA, TC; cannam@167: { cannam@167: E Tp, Tw, Ty, Tz; cannam@167: Tp = Tl - To; cannam@167: Tw = Ts - Tv; cannam@167: Tx = FMA(KP618033988, Tw, Tp); cannam@167: TB = FNMS(KP618033988, Tp, Tw); cannam@167: Ty = FMA(KP250000000, Tf, Ti); cannam@167: Tz = Te - T7; cannam@167: TA = FMA(KP559016994, Tz, Ty); cannam@167: TC = FNMS(KP559016994, Tz, Ty); cannam@167: } cannam@167: Ci[WS(csi, 1)] = -(KP951056516 * (FNMS(KP910592997, TA, Tx))); cannam@167: Ci[WS(csi, 7)] = KP951056516 * (FMA(KP910592997, TC, TB)); cannam@167: Ci[WS(csi, 4)] = KP951056516 * (FMA(KP910592997, TA, Tx)); cannam@167: Ci[WS(csi, 2)] = KP951056516 * (FNMS(KP910592997, TC, TB)); cannam@167: } cannam@167: } cannam@167: } cannam@167: } cannam@167: } cannam@167: cannam@167: static const kr2c_desc desc = { 15, "r2cf_15", {36, 7, 28, 0}, &GENUS }; cannam@167: cannam@167: void X(codelet_r2cf_15) (planner *p) { cannam@167: X(kr2c_register) (p, r2cf_15, &desc); cannam@167: } cannam@167: cannam@167: #else cannam@167: cannam@167: /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 15 -name r2cf_15 -include rdft/scalar/r2cf.h */ cannam@167: cannam@167: /* cannam@167: * This function contains 64 FP additions, 25 FP multiplications, cannam@167: * (or, 50 additions, 11 multiplications, 14 fused multiply/add), cannam@167: * 47 stack variables, 10 constants, and 30 memory accesses cannam@167: */ cannam@167: #include "rdft/scalar/r2cf.h" cannam@167: cannam@167: static void r2cf_15(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(KP484122918, +0.484122918275927110647408174972799951354115213); cannam@167: DK(KP216506350, +0.216506350946109661690930792688234045867850657); cannam@167: DK(KP951056516, +0.951056516295153572116439333379382143405698634); cannam@167: DK(KP587785252, +0.587785252292473129168705954639072768597652438); cannam@167: DK(KP250000000, +0.250000000000000000000000000000000000000000000); cannam@167: DK(KP559016994, +0.559016994374947424102293417182819058860154590); cannam@167: DK(KP509036960, +0.509036960455127183450980863393907648510733164); cannam@167: DK(KP823639103, +0.823639103546331925877420039278190003029660514); cannam@167: DK(KP866025403, +0.866025403784438646763723170752936183471402627); cannam@167: DK(KP500000000, +0.500000000000000000000000000000000000000000000); cannam@167: { cannam@167: INT i; cannam@167: for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(60, rs), MAKE_VOLATILE_STRIDE(60, csr), MAKE_VOLATILE_STRIDE(60, csi)) { cannam@167: E Ti, TR, TL, TD, TE, T7, Te, Tf, TV, TW, TX, Tv, Ty, TH, To; cannam@167: E Tr, TG, TS, TT, TU; cannam@167: { cannam@167: E TJ, Tg, Th, TK; cannam@167: TJ = R0[0]; cannam@167: Tg = R0[WS(rs, 5)]; cannam@167: Th = R1[WS(rs, 2)]; cannam@167: TK = Th + Tg; cannam@167: Ti = Tg - Th; cannam@167: TR = TJ + TK; cannam@167: TL = FNMS(KP500000000, TK, TJ); cannam@167: } cannam@167: { cannam@167: E Tm, Tt, Tw, Tp, T3, Tx, Ta, Tn, Td, Tq, T6, Tu; cannam@167: Tm = R1[WS(rs, 1)]; cannam@167: Tt = R0[WS(rs, 3)]; cannam@167: Tw = R1[WS(rs, 4)]; cannam@167: Tp = R0[WS(rs, 6)]; cannam@167: { cannam@167: E T1, T2, T8, T9; cannam@167: T1 = R0[WS(rs, 7)]; cannam@167: T2 = R0[WS(rs, 2)]; cannam@167: T3 = T1 - T2; cannam@167: Tx = T1 + T2; cannam@167: T8 = R1[WS(rs, 6)]; cannam@167: T9 = R0[WS(rs, 4)]; cannam@167: Ta = T8 - T9; cannam@167: Tn = T9 + T8; cannam@167: } cannam@167: { cannam@167: E Tb, Tc, T4, T5; cannam@167: Tb = R1[WS(rs, 3)]; cannam@167: Tc = R0[WS(rs, 1)]; cannam@167: Td = Tb - Tc; cannam@167: Tq = Tc + Tb; cannam@167: T4 = R1[0]; cannam@167: T5 = R1[WS(rs, 5)]; cannam@167: T6 = T4 - T5; cannam@167: Tu = T5 + T4; cannam@167: } cannam@167: TD = Ta - Td; cannam@167: TE = T6 + T3; cannam@167: T7 = T3 - T6; cannam@167: Te = Ta + Td; cannam@167: Tf = T7 - Te; cannam@167: TV = Tt + Tu; cannam@167: TW = Tw + Tx; cannam@167: TX = TV + TW; cannam@167: Tv = FNMS(KP500000000, Tu, Tt); cannam@167: Ty = FNMS(KP500000000, Tx, Tw); cannam@167: TH = Tv + Ty; cannam@167: To = FNMS(KP500000000, Tn, Tm); cannam@167: Tr = FNMS(KP500000000, Tq, Tp); cannam@167: TG = To + Tr; cannam@167: TS = Tm + Tn; cannam@167: TT = Tp + Tq; cannam@167: TU = TS + TT; cannam@167: } cannam@167: Ci[WS(csi, 5)] = KP866025403 * (Tf - Ti); cannam@167: { cannam@167: E TF, TP, TI, TM, TN, TQ, TO; cannam@167: TF = FMA(KP823639103, TD, KP509036960 * TE); cannam@167: TP = FNMS(KP509036960, TD, KP823639103 * TE); cannam@167: TI = KP559016994 * (TG - TH); cannam@167: TM = TG + TH; cannam@167: TN = FNMS(KP250000000, TM, TL); cannam@167: Cr[WS(csr, 5)] = TL + TM; cannam@167: TQ = TN - TI; cannam@167: Cr[WS(csr, 2)] = TP + TQ; cannam@167: Cr[WS(csr, 7)] = TQ - TP; cannam@167: TO = TI + TN; cannam@167: Cr[WS(csr, 1)] = TF + TO; cannam@167: Cr[WS(csr, 4)] = TO - TF; cannam@167: } cannam@167: { cannam@167: E T11, T12, T10, TY, TZ; cannam@167: T11 = TS - TT; cannam@167: T12 = TW - TV; cannam@167: Ci[WS(csi, 3)] = FMA(KP587785252, T11, KP951056516 * T12); cannam@167: Ci[WS(csi, 6)] = FNMS(KP951056516, T11, KP587785252 * T12); cannam@167: T10 = KP559016994 * (TU - TX); cannam@167: TY = TU + TX; cannam@167: TZ = FNMS(KP250000000, TY, TR); cannam@167: Cr[WS(csr, 3)] = TZ - T10; cannam@167: Cr[0] = TR + TY; cannam@167: Cr[WS(csr, 6)] = T10 + TZ; cannam@167: { cannam@167: E Tl, TB, TA, TC; cannam@167: { cannam@167: E Tj, Tk, Ts, Tz; cannam@167: Tj = FMA(KP866025403, Ti, KP216506350 * Tf); cannam@167: Tk = KP484122918 * (Te + T7); cannam@167: Tl = Tj + Tk; cannam@167: TB = Tk - Tj; cannam@167: Ts = To - Tr; cannam@167: Tz = Tv - Ty; cannam@167: TA = FMA(KP951056516, Ts, KP587785252 * Tz); cannam@167: TC = FNMS(KP587785252, Ts, KP951056516 * Tz); cannam@167: } cannam@167: Ci[WS(csi, 1)] = Tl - TA; cannam@167: Ci[WS(csi, 7)] = TC - TB; cannam@167: Ci[WS(csi, 4)] = Tl + TA; cannam@167: Ci[WS(csi, 2)] = TB + TC; cannam@167: } cannam@167: } cannam@167: } cannam@167: } cannam@167: } cannam@167: cannam@167: static const kr2c_desc desc = { 15, "r2cf_15", {50, 11, 14, 0}, &GENUS }; cannam@167: cannam@167: void X(codelet_r2cf_15) (planner *p) { cannam@167: X(kr2c_register) (p, r2cf_15, &desc); cannam@167: } cannam@167: cannam@167: #endif