Chris@10: /* Chris@10: * Copyright (c) 2003, 2007-11 Matteo Frigo Chris@10: * Copyright (c) 2003, 2007-11 Massachusetts Institute of Technology Chris@10: * Chris@10: * This program is free software; you can redistribute it and/or modify Chris@10: * it under the terms of the GNU General Public License as published by Chris@10: * the Free Software Foundation; either version 2 of the License, or Chris@10: * (at your option) any later version. Chris@10: * Chris@10: * This program is distributed in the hope that it will be useful, Chris@10: * but WITHOUT ANY WARRANTY; without even the implied warranty of Chris@10: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the Chris@10: * GNU General Public License for more details. Chris@10: * Chris@10: * You should have received a copy of the GNU General Public License Chris@10: * along with this program; if not, write to the Free Software Chris@10: * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA Chris@10: * Chris@10: */ Chris@10: Chris@10: /* This file was automatically generated --- DO NOT EDIT */ Chris@10: /* Generated on Sun Nov 25 07:41:08 EST 2012 */ Chris@10: Chris@10: #include "codelet-rdft.h" Chris@10: Chris@10: #ifdef HAVE_FMA Chris@10: Chris@10: /* Generated by: ../../../genfft/gen_r2cb.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -sign 1 -n 13 -name r2cb_13 -include r2cb.h */ Chris@10: Chris@10: /* Chris@10: * This function contains 76 FP additions, 58 FP multiplications, Chris@10: * (or, 18 additions, 0 multiplications, 58 fused multiply/add), Chris@10: * 76 stack variables, 26 constants, and 26 memory accesses Chris@10: */ Chris@10: #include "r2cb.h" Chris@10: Chris@10: static void r2cb_13(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs) Chris@10: { Chris@10: DK(KP968287244, +0.968287244361984016049539446938120421179794516); Chris@10: DK(KP875502302, +0.875502302409147941146295545768755143177842006); Chris@10: DK(KP1_150281458, +1.150281458948006242736771094910906776922003215); Chris@10: DK(KP1_040057143, +1.040057143777729238234261000998465604986476278); Chris@10: DK(KP1_200954543, +1.200954543865330565851538506669526018704025697); Chris@10: DK(KP769338817, +0.769338817572980603471413688209101117038278899); Chris@10: DK(KP600925212, +0.600925212577331548853203544578415991041882762); Chris@10: DK(KP1_033041561, +1.033041561246979445681802577138034271410067244); Chris@10: DK(KP1_007074065, +1.007074065727533254493747707736933954186697125); Chris@10: DK(KP503537032, +0.503537032863766627246873853868466977093348562); Chris@10: DK(KP581704778, +0.581704778510515730456870384989698884939833902); Chris@10: DK(KP859542535, +0.859542535098774820163672132761689612766401925); Chris@10: DK(KP166666666, +0.166666666666666666666666666666666666666666667); Chris@10: DK(KP2_000000000, +2.000000000000000000000000000000000000000000000); Chris@10: DK(KP301479260, +0.301479260047709873958013540496673347309208464); Chris@10: DK(KP226109445, +0.226109445035782405468510155372505010481906348); Chris@10: DK(KP686558370, +0.686558370781754340655719594850823015421401653); Chris@10: DK(KP514918778, +0.514918778086315755491789696138117261566051239); Chris@10: DK(KP957805992, +0.957805992594665126462521754605754580515587217); Chris@10: DK(KP522026385, +0.522026385161275033714027226654165028300441940); Chris@10: DK(KP853480001, +0.853480001859823990758994934970528322872359049); Chris@10: DK(KP038632954, +0.038632954644348171955506895830342264440241080); Chris@10: DK(KP612264650, +0.612264650376756543746494474777125408779395514); Chris@10: DK(KP302775637, +0.302775637731994646559610633735247973125648287); Chris@10: DK(KP866025403, +0.866025403784438646763723170752936183471402627); Chris@10: DK(KP500000000, +0.500000000000000000000000000000000000000000000); Chris@10: { Chris@10: INT i; Chris@10: for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(52, rs), MAKE_VOLATILE_STRIDE(52, csr), MAKE_VOLATILE_STRIDE(52, csi)) { Chris@10: E TW, T14, TS, TO, T18, T1e, TY, TX, TQ, Tq, TP, Tl, T1d, Tr; Chris@10: { Chris@10: E T1, TN, T16, TJ, TV, TG, TU, Tf, T2, T3, Tb, Ti, T4; Chris@10: { Chris@10: E Ts, TB, Tx, Ty, Tv, TE, Tt, Tu, Tz, TC; Chris@10: Ts = Ci[WS(csi, 5)]; Chris@10: Tt = Ci[WS(csi, 2)]; Chris@10: Tu = Ci[WS(csi, 6)]; Chris@10: TB = Ci[WS(csi, 1)]; Chris@10: Tx = Ci[WS(csi, 3)]; Chris@10: Ty = Ci[WS(csi, 4)]; Chris@10: Tv = Tt + Tu; Chris@10: TE = Tu - Tt; Chris@10: T1 = Cr[0]; Chris@10: Tz = Tx + Ty; Chris@10: TC = Tx - Ty; Chris@10: { Chris@10: E TL, Tw, T7, Ta; Chris@10: TL = Ts + Tv; Chris@10: Tw = FNMS(KP500000000, Tv, Ts); Chris@10: T7 = Cr[WS(csr, 5)]; Chris@10: { Chris@10: E TD, TM, TA, TH; Chris@10: TD = FNMS(KP500000000, TC, TB); Chris@10: TM = TB + TC; Chris@10: TA = FMA(KP866025403, Tz, Tw); Chris@10: TH = FNMS(KP866025403, Tz, Tw); Chris@10: TN = FMA(KP302775637, TM, TL); Chris@10: T16 = FNMS(KP302775637, TL, TM); Chris@10: { Chris@10: E TF, TI, T8, T9; Chris@10: TF = FMA(KP866025403, TE, TD); Chris@10: TI = FNMS(KP866025403, TE, TD); Chris@10: T8 = Cr[WS(csr, 2)]; Chris@10: T9 = Cr[WS(csr, 6)]; Chris@10: TJ = FNMS(KP612264650, TI, TH); Chris@10: TV = FMA(KP612264650, TH, TI); Chris@10: TG = FNMS(KP038632954, TF, TA); Chris@10: TU = FMA(KP038632954, TA, TF); Chris@10: Tf = T8 - T9; Chris@10: Ta = T8 + T9; Chris@10: } Chris@10: } Chris@10: T2 = Cr[WS(csr, 1)]; Chris@10: T3 = Cr[WS(csr, 3)]; Chris@10: Tb = T7 + Ta; Chris@10: Ti = FMS(KP500000000, Ta, T7); Chris@10: T4 = Cr[WS(csr, 4)]; Chris@10: } Chris@10: } Chris@10: { Chris@10: E T17, TK, T5, Te, Tk, Td; Chris@10: TW = FMA(KP853480001, TV, TU); Chris@10: T17 = FNMS(KP853480001, TV, TU); Chris@10: TK = FNMS(KP853480001, TJ, TG); Chris@10: T14 = FMA(KP853480001, TJ, TG); Chris@10: T5 = T3 + T4; Chris@10: Te = T3 - T4; Chris@10: { Chris@10: E Tn, Tg, Th, T6; Chris@10: TS = FNMS(KP522026385, TK, TN); Chris@10: TO = FMA(KP957805992, TN, TK); Chris@10: Tn = Te - Tf; Chris@10: Tg = Te + Tf; Chris@10: Th = FNMS(KP500000000, T5, T2); Chris@10: T6 = T2 + T5; Chris@10: T18 = FNMS(KP522026385, T17, T16); Chris@10: T1e = FMA(KP957805992, T16, T17); Chris@10: { Chris@10: E Tm, Tj, Tc, Tp, To; Chris@10: Tm = Th + Ti; Chris@10: Tj = Th - Ti; Chris@10: Tc = T6 + Tb; Chris@10: Tp = T6 - Tb; Chris@10: To = FNMS(KP514918778, Tn, Tm); Chris@10: TY = FMA(KP686558370, Tm, Tn); Chris@10: TX = FNMS(KP226109445, Tg, Tj); Chris@10: Tk = FMA(KP301479260, Tj, Tg); Chris@10: R0[0] = FMA(KP2_000000000, Tc, T1); Chris@10: Td = FNMS(KP166666666, Tc, T1); Chris@10: TQ = FNMS(KP859542535, To, Tp); Chris@10: Tq = FMA(KP581704778, Tp, To); Chris@10: } Chris@10: } Chris@10: TP = FNMS(KP503537032, Tk, Td); Chris@10: Tl = FMA(KP1_007074065, Tk, Td); Chris@10: } Chris@10: } Chris@10: T1d = FNMS(KP1_033041561, Tq, Tl); Chris@10: Tr = FMA(KP1_033041561, Tq, Tl); Chris@10: { Chris@10: E T13, TR, T19, TZ; Chris@10: T13 = FNMS(KP600925212, TQ, TP); Chris@10: TR = FMA(KP600925212, TQ, TP); Chris@10: T19 = FMA(KP769338817, TY, TX); Chris@10: TZ = FNMS(KP769338817, TY, TX); Chris@10: R0[WS(rs, 4)] = FMA(KP1_200954543, T1e, T1d); Chris@10: R1[WS(rs, 2)] = FNMS(KP1_200954543, T1e, T1d); Chris@10: R0[WS(rs, 6)] = FMA(KP1_200954543, TO, Tr); Chris@10: R1[0] = FNMS(KP1_200954543, TO, Tr); Chris@10: { Chris@10: E T1b, T15, T11, TT; Chris@10: T1b = FNMS(KP1_040057143, T14, T13); Chris@10: T15 = FMA(KP1_040057143, T14, T13); Chris@10: T11 = FMA(KP1_150281458, TS, TR); Chris@10: TT = FNMS(KP1_150281458, TS, TR); Chris@10: { Chris@10: E T1c, T1a, T12, T10; Chris@10: T1c = FMA(KP875502302, T19, T18); Chris@10: T1a = FNMS(KP875502302, T19, T18); Chris@10: T12 = FMA(KP968287244, TZ, TW); Chris@10: T10 = FNMS(KP968287244, TZ, TW); Chris@10: R1[WS(rs, 5)] = FMA(KP1_150281458, T1c, T1b); Chris@10: R0[WS(rs, 3)] = FNMS(KP1_150281458, T1c, T1b); Chris@10: R1[WS(rs, 3)] = FMA(KP1_150281458, T1a, T15); Chris@10: R0[WS(rs, 1)] = FNMS(KP1_150281458, T1a, T15); Chris@10: R0[WS(rs, 5)] = FMA(KP1_040057143, T12, T11); Chris@10: R0[WS(rs, 2)] = FNMS(KP1_040057143, T12, T11); Chris@10: R1[WS(rs, 4)] = FMA(KP1_040057143, T10, TT); Chris@10: R1[WS(rs, 1)] = FNMS(KP1_040057143, T10, TT); Chris@10: } Chris@10: } Chris@10: } Chris@10: } Chris@10: } Chris@10: } Chris@10: Chris@10: static const kr2c_desc desc = { 13, "r2cb_13", {18, 0, 58, 0}, &GENUS }; Chris@10: Chris@10: void X(codelet_r2cb_13) (planner *p) { Chris@10: X(kr2c_register) (p, r2cb_13, &desc); Chris@10: } Chris@10: Chris@10: #else /* HAVE_FMA */ Chris@10: Chris@10: /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 13 -name r2cb_13 -include r2cb.h */ Chris@10: Chris@10: /* Chris@10: * This function contains 76 FP additions, 35 FP multiplications, Chris@10: * (or, 56 additions, 15 multiplications, 20 fused multiply/add), Chris@10: * 56 stack variables, 19 constants, and 26 memory accesses Chris@10: */ Chris@10: #include "r2cb.h" Chris@10: Chris@10: static void r2cb_13(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs) Chris@10: { Chris@10: DK(KP1_007074065, +1.007074065727533254493747707736933954186697125); Chris@10: DK(KP227708958, +0.227708958111581597949308691735310621069285120); Chris@10: DK(KP531932498, +0.531932498429674575175042127684371897596660533); Chris@10: DK(KP774781170, +0.774781170935234584261351932853525703557550433); Chris@10: DK(KP265966249, +0.265966249214837287587521063842185948798330267); Chris@10: DK(KP516520780, +0.516520780623489722840901288569017135705033622); Chris@10: DK(KP151805972, +0.151805972074387731966205794490207080712856746); Chris@10: DK(KP503537032, +0.503537032863766627246873853868466977093348562); Chris@10: DK(KP166666666, +0.166666666666666666666666666666666666666666667); Chris@10: DK(KP600925212, +0.600925212577331548853203544578415991041882762); Chris@10: DK(KP500000000, +0.500000000000000000000000000000000000000000000); Chris@10: DK(KP256247671, +0.256247671582936600958684654061725059144125175); Chris@10: DK(KP156891391, +0.156891391051584611046832726756003269660212636); Chris@10: DK(KP348277202, +0.348277202304271810011321589858529485233929352); Chris@10: DK(KP1_150281458, +1.150281458948006242736771094910906776922003215); Chris@10: DK(KP300238635, +0.300238635966332641462884626667381504676006424); Chris@10: DK(KP011599105, +0.011599105605768290721655456654083252189827041); Chris@10: DK(KP1_732050807, +1.732050807568877293527446341505872366942805254); Chris@10: DK(KP2_000000000, +2.000000000000000000000000000000000000000000000); Chris@10: { Chris@10: INT i; Chris@10: for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(52, rs), MAKE_VOLATILE_STRIDE(52, csr), MAKE_VOLATILE_STRIDE(52, csi)) { Chris@10: E TG, TS, TR, T15, TJ, TT, T1, Tm, Tc, Td, Tg, Tj, Tk, Tn, To; Chris@10: E Tp; Chris@10: { Chris@10: E Ts, Tv, Tw, TE, TC, TB, Tz, TD, TA, TF; Chris@10: { Chris@10: E Tt, Tu, Tx, Ty; Chris@10: Ts = Ci[WS(csi, 1)]; Chris@10: Tt = Ci[WS(csi, 3)]; Chris@10: Tu = Ci[WS(csi, 4)]; Chris@10: Tv = Tt - Tu; Chris@10: Tw = FMS(KP2_000000000, Ts, Tv); Chris@10: TE = KP1_732050807 * (Tt + Tu); Chris@10: TC = Ci[WS(csi, 5)]; Chris@10: Tx = Ci[WS(csi, 6)]; Chris@10: Ty = Ci[WS(csi, 2)]; Chris@10: TB = Tx + Ty; Chris@10: Tz = KP1_732050807 * (Tx - Ty); Chris@10: TD = FNMS(KP2_000000000, TC, TB); Chris@10: } Chris@10: TA = Tw + Tz; Chris@10: TF = TD - TE; Chris@10: TG = FMA(KP011599105, TA, KP300238635 * TF); Chris@10: TS = FNMS(KP011599105, TF, KP300238635 * TA); Chris@10: { Chris@10: E TP, TQ, TH, TI; Chris@10: TP = Ts + Tv; Chris@10: TQ = TB + TC; Chris@10: TR = FNMS(KP348277202, TQ, KP1_150281458 * TP); Chris@10: T15 = FMA(KP348277202, TP, KP1_150281458 * TQ); Chris@10: TH = Tw - Tz; Chris@10: TI = TE + TD; Chris@10: TJ = FMA(KP156891391, TH, KP256247671 * TI); Chris@10: TT = FNMS(KP256247671, TH, KP156891391 * TI); Chris@10: } Chris@10: } Chris@10: { Chris@10: E Tb, Ti, Tf, T6, Th, Te; Chris@10: T1 = Cr[0]; Chris@10: { Chris@10: E T7, T8, T9, Ta; Chris@10: T7 = Cr[WS(csr, 5)]; Chris@10: T8 = Cr[WS(csr, 2)]; Chris@10: T9 = Cr[WS(csr, 6)]; Chris@10: Ta = T8 + T9; Chris@10: Tb = T7 + Ta; Chris@10: Ti = FNMS(KP500000000, Ta, T7); Chris@10: Tf = T8 - T9; Chris@10: } Chris@10: { Chris@10: E T2, T3, T4, T5; Chris@10: T2 = Cr[WS(csr, 1)]; Chris@10: T3 = Cr[WS(csr, 3)]; Chris@10: T4 = Cr[WS(csr, 4)]; Chris@10: T5 = T3 + T4; Chris@10: T6 = T2 + T5; Chris@10: Th = FNMS(KP500000000, T5, T2); Chris@10: Te = T3 - T4; Chris@10: } Chris@10: Tm = KP600925212 * (T6 - Tb); Chris@10: Tc = T6 + Tb; Chris@10: Td = FNMS(KP166666666, Tc, T1); Chris@10: Tg = Te + Tf; Chris@10: Tj = Th + Ti; Chris@10: Tk = FMA(KP503537032, Tg, KP151805972 * Tj); Chris@10: Tn = Th - Ti; Chris@10: To = Te - Tf; Chris@10: Tp = FNMS(KP265966249, To, KP516520780 * Tn); Chris@10: } Chris@10: R0[0] = FMA(KP2_000000000, Tc, T1); Chris@10: { Chris@10: E TK, T1b, TV, T12, T16, T18, TO, T1a, Tr, T17, T11, T13; Chris@10: { Chris@10: E TU, T14, TM, TN; Chris@10: TK = KP1_732050807 * (TG + TJ); Chris@10: T1b = KP1_732050807 * (TS - TT); Chris@10: TU = TS + TT; Chris@10: TV = TR - TU; Chris@10: T12 = FMA(KP2_000000000, TU, TR); Chris@10: T14 = TG - TJ; Chris@10: T16 = FMS(KP2_000000000, T14, T15); Chris@10: T18 = T14 + T15; Chris@10: TM = FMA(KP774781170, To, KP531932498 * Tn); Chris@10: TN = FNMS(KP1_007074065, Tj, KP227708958 * Tg); Chris@10: TO = TM - TN; Chris@10: T1a = TM + TN; Chris@10: { Chris@10: E Tl, Tq, TZ, T10; Chris@10: Tl = Td - Tk; Chris@10: Tq = Tm - Tp; Chris@10: Tr = Tl - Tq; Chris@10: T17 = Tq + Tl; Chris@10: TZ = FMA(KP2_000000000, Tk, Td); Chris@10: T10 = FMA(KP2_000000000, Tp, Tm); Chris@10: T11 = TZ - T10; Chris@10: T13 = T10 + TZ; Chris@10: } Chris@10: } Chris@10: R1[WS(rs, 2)] = T11 - T12; Chris@10: R0[WS(rs, 6)] = T13 - T16; Chris@10: R1[0] = T13 + T16; Chris@10: R0[WS(rs, 4)] = T11 + T12; Chris@10: { Chris@10: E TL, TW, T19, T1c; Chris@10: TL = Tr - TK; Chris@10: TW = TO - TV; Chris@10: R1[WS(rs, 3)] = TL - TW; Chris@10: R0[WS(rs, 1)] = TL + TW; Chris@10: T19 = T17 - T18; Chris@10: T1c = T1a + T1b; Chris@10: R1[WS(rs, 1)] = T19 - T1c; Chris@10: R1[WS(rs, 4)] = T1c + T19; Chris@10: } Chris@10: { Chris@10: E T1d, T1e, TX, TY; Chris@10: T1d = T1a - T1b; Chris@10: T1e = T17 + T18; Chris@10: R0[WS(rs, 2)] = T1d + T1e; Chris@10: R0[WS(rs, 5)] = T1e - T1d; Chris@10: TX = Tr + TK; Chris@10: TY = TO + TV; Chris@10: R0[WS(rs, 3)] = TX - TY; Chris@10: R1[WS(rs, 5)] = TX + TY; Chris@10: } Chris@10: } Chris@10: } Chris@10: } Chris@10: } Chris@10: Chris@10: static const kr2c_desc desc = { 13, "r2cb_13", {56, 15, 20, 0}, &GENUS }; Chris@10: Chris@10: void X(codelet_r2cb_13) (planner *p) { Chris@10: X(kr2c_register) (p, r2cb_13, &desc); Chris@10: } Chris@10: Chris@10: #endif /* HAVE_FMA */