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 07:40:14 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_r2cf.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 9 -name r2cfII_9 -dft-II -include r2cfII.h */ Chris@10: Chris@10: /* Chris@10: * This function contains 42 FP additions, 34 FP multiplications, Chris@10: * (or, 12 additions, 4 multiplications, 30 fused multiply/add), Chris@10: * 46 stack variables, 17 constants, and 18 memory accesses Chris@10: */ Chris@10: #include "r2cfII.h" Chris@10: Chris@10: static void r2cfII_9(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(KP939692620, +0.939692620785908384054109277324731469936208134); Chris@10: DK(KP879385241, +0.879385241571816768108218554649462939872416269); Chris@10: DK(KP984807753, +0.984807753012208059366743024589523013670643252); Chris@10: DK(KP852868531, +0.852868531952443209628250963940074071936020296); Chris@10: DK(KP666666666, +0.666666666666666666666666666666666666666666667); Chris@10: DK(KP673648177, +0.673648177666930348851716626769314796000375677); Chris@10: DK(KP898197570, +0.898197570222573798468955502359086394667167570); Chris@10: DK(KP826351822, +0.826351822333069651148283373230685203999624323); Chris@10: DK(KP907603734, +0.907603734547952313649323976213898122064543220); Chris@10: DK(KP866025403, +0.866025403784438646763723170752936183471402627); Chris@10: DK(KP420276625, +0.420276625461206169731530603237061658838781920); Chris@10: DK(KP315207469, +0.315207469095904627298647952427796244129086440); Chris@10: DK(KP203604859, +0.203604859554852403062088995281827210665664861); Chris@10: DK(KP152703644, +0.152703644666139302296566746461370407999248646); Chris@10: DK(KP726681596, +0.726681596905677465811651808188092531873167623); Chris@10: DK(KP968908795, +0.968908795874236621082202410917456709164223497); Chris@10: DK(KP500000000, +0.500000000000000000000000000000000000000000000); Chris@10: { Chris@10: INT i; Chris@10: for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(36, rs), MAKE_VOLATILE_STRIDE(36, csr), MAKE_VOLATILE_STRIDE(36, csi)) { Chris@10: E To, T5, Tp, Ta, Ti, Tm, TB, Tq, Tt, Tf, Th; Chris@10: { Chris@10: E T1, T6, T4, Tb, Tk, T9, Tc, Td, Tl, Te; Chris@10: { Chris@10: E T2, T3, T7, T8; Chris@10: T1 = R0[0]; Chris@10: T2 = R0[WS(rs, 3)]; Chris@10: T3 = R1[WS(rs, 1)]; Chris@10: T6 = R0[WS(rs, 1)]; Chris@10: T7 = R0[WS(rs, 4)]; Chris@10: T8 = R1[WS(rs, 2)]; Chris@10: T4 = T2 - T3; Chris@10: To = T2 + T3; Chris@10: Tb = R0[WS(rs, 2)]; Chris@10: Tk = T7 + T8; Chris@10: T9 = T7 - T8; Chris@10: Tc = R1[0]; Chris@10: Td = R1[WS(rs, 3)]; Chris@10: } Chris@10: T5 = T1 + T4; Chris@10: Tp = FNMS(KP500000000, T4, T1); Chris@10: Ta = T6 + T9; Chris@10: Tl = FNMS(KP500000000, T9, T6); Chris@10: Te = Tc + Td; Chris@10: Ti = Tc - Td; Chris@10: Tm = FMA(KP968908795, Tl, Tk); Chris@10: TB = FNMS(KP726681596, Tk, Tl); Chris@10: Tq = FNMS(KP152703644, Tk, Tl); Chris@10: Tt = FMA(KP203604859, Tl, Tk); Chris@10: Tf = Tb - Te; Chris@10: Th = FMA(KP500000000, Te, Tb); Chris@10: } Chris@10: { Chris@10: E Ts, Tr, TA, Tj, Tg; Chris@10: Ts = FMA(KP315207469, Ti, Th); Chris@10: Tr = FNMS(KP420276625, Th, Ti); Chris@10: TA = FMA(KP203604859, Th, Ti); Chris@10: Tj = FNMS(KP152703644, Ti, Th); Chris@10: Tg = Ta + Tf; Chris@10: Ci[WS(csi, 1)] = KP866025403 * (Tf - Ta); Chris@10: { Chris@10: E Tu, Tx, TF, TC; Chris@10: Tu = FNMS(KP907603734, Tt, Ts); Chris@10: Tx = FNMS(KP826351822, Tr, Tq); Chris@10: TF = FMA(KP898197570, TB, TA); Chris@10: TC = FNMS(KP898197570, TB, TA); Chris@10: { Chris@10: E TE, Tn, Tv, Ty; Chris@10: TE = FNMS(KP673648177, Tm, Tj); Chris@10: Tn = FMA(KP673648177, Tm, Tj); Chris@10: Cr[WS(csr, 4)] = T5 + Tg; Chris@10: Cr[WS(csr, 1)] = FNMS(KP500000000, Tg, T5); Chris@10: Tv = FNMS(KP666666666, Tu, Tr); Chris@10: Ty = FNMS(KP666666666, Tx, Tt); Chris@10: Cr[0] = FMA(KP852868531, TF, Tp); Chris@10: { Chris@10: E TG, TD, Tw, Tz; Chris@10: TG = FMA(KP500000000, TF, TE); Chris@10: Ci[0] = -(KP984807753 * (FMA(KP879385241, To, Tn))); Chris@10: TD = FNMS(KP666666666, Tn, TC); Chris@10: Tw = FMA(KP826351822, Tv, Tq); Chris@10: Tz = FMA(KP907603734, Ty, Ts); Chris@10: Cr[WS(csr, 3)] = FNMS(KP852868531, TG, Tp); Chris@10: Ci[WS(csi, 3)] = -(KP866025403 * (FMA(KP852868531, TD, To))); Chris@10: Cr[WS(csr, 2)] = FNMS(KP852868531, Tw, Tp); Chris@10: Ci[WS(csi, 2)] = KP866025403 * (FNMS(KP939692620, Tz, To)); Chris@10: } Chris@10: } Chris@10: } Chris@10: } Chris@10: } Chris@10: } Chris@10: } Chris@10: Chris@10: static const kr2c_desc desc = { 9, "r2cfII_9", {12, 4, 30, 0}, &GENUS }; Chris@10: Chris@10: void X(codelet_r2cfII_9) (planner *p) { Chris@10: X(kr2c_register) (p, r2cfII_9, &desc); Chris@10: } Chris@10: Chris@10: #else /* HAVE_FMA */ Chris@10: Chris@10: /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 9 -name r2cfII_9 -dft-II -include r2cfII.h */ Chris@10: Chris@10: /* Chris@10: * This function contains 42 FP additions, 30 FP multiplications, Chris@10: * (or, 25 additions, 13 multiplications, 17 fused multiply/add), Chris@10: * 39 stack variables, 14 constants, and 18 memory accesses Chris@10: */ Chris@10: #include "r2cfII.h" Chris@10: Chris@10: static void r2cfII_9(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(KP663413948, +0.663413948168938396205421319635891297216863310); Chris@10: DK(KP642787609, +0.642787609686539326322643409907263432907559884); Chris@10: DK(KP556670399, +0.556670399226419366452912952047023132968291906); Chris@10: DK(KP766044443, +0.766044443118978035202392650555416673935832457); Chris@10: DK(KP852868531, +0.852868531952443209628250963940074071936020296); Chris@10: DK(KP173648177, +0.173648177666930348851716626769314796000375677); Chris@10: DK(KP984807753, +0.984807753012208059366743024589523013670643252); Chris@10: DK(KP150383733, +0.150383733180435296639271897612501926072238258); Chris@10: DK(KP813797681, +0.813797681349373692844693217248393223289101568); Chris@10: DK(KP342020143, +0.342020143325668733044099614682259580763083368); Chris@10: DK(KP939692620, +0.939692620785908384054109277324731469936208134); Chris@10: DK(KP296198132, +0.296198132726023843175338011893050938967728390); 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 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(36, rs), MAKE_VOLATILE_STRIDE(36, csr), MAKE_VOLATILE_STRIDE(36, csi)) { Chris@10: E T1, T4, To, Ta, Tl, Tk, Tf, Ti, Th, T2, T3, T5, Tg; Chris@10: T1 = R0[0]; Chris@10: T2 = R1[WS(rs, 1)]; Chris@10: T3 = R0[WS(rs, 3)]; Chris@10: T4 = T2 - T3; Chris@10: To = T2 + T3; Chris@10: { Chris@10: E T6, T7, T8, T9; Chris@10: T6 = R0[WS(rs, 1)]; Chris@10: T7 = R1[WS(rs, 2)]; Chris@10: T8 = R0[WS(rs, 4)]; Chris@10: T9 = T7 - T8; Chris@10: Ta = T6 - T9; Chris@10: Tl = T7 + T8; Chris@10: Tk = FMA(KP500000000, T9, T6); Chris@10: } Chris@10: { Chris@10: E Tb, Tc, Td, Te; Chris@10: Tb = R0[WS(rs, 2)]; Chris@10: Tc = R1[0]; Chris@10: Td = R1[WS(rs, 3)]; Chris@10: Te = Tc + Td; Chris@10: Tf = Tb - Te; Chris@10: Ti = FMA(KP500000000, Te, Tb); Chris@10: Th = Tc - Td; Chris@10: } Chris@10: Ci[WS(csi, 1)] = KP866025403 * (Tf - Ta); Chris@10: T5 = T1 - T4; Chris@10: Tg = Ta + Tf; Chris@10: Cr[WS(csr, 1)] = FNMS(KP500000000, Tg, T5); Chris@10: Cr[WS(csr, 4)] = T5 + Tg; Chris@10: { Chris@10: E Tr, Tt, Tw, Tv, Tu, Tp, Tq, Ts, Tj, Tm, Tn; Chris@10: Tr = FMA(KP500000000, T4, T1); Chris@10: Tt = FMA(KP296198132, Th, KP939692620 * Ti); Chris@10: Tw = FNMS(KP813797681, Th, KP342020143 * Ti); Chris@10: Tv = FNMS(KP984807753, Tk, KP150383733 * Tl); Chris@10: Tu = FMA(KP173648177, Tk, KP852868531 * Tl); Chris@10: Tp = FNMS(KP556670399, Tl, KP766044443 * Tk); Chris@10: Tq = FMA(KP852868531, Th, KP173648177 * Ti); Chris@10: Ts = Tp + Tq; Chris@10: Tj = FNMS(KP984807753, Ti, KP150383733 * Th); Chris@10: Tm = FMA(KP642787609, Tk, KP663413948 * Tl); Chris@10: Tn = Tj - Tm; Chris@10: Ci[0] = FNMS(KP866025403, To, Tn); Chris@10: Cr[0] = Tr + Ts; Chris@10: Ci[WS(csi, 3)] = FNMS(KP500000000, Tn, KP866025403 * ((Tp - Tq) - To)); Chris@10: Cr[WS(csr, 3)] = FMA(KP866025403, Tm + Tj, Tr) - (KP500000000 * Ts); Chris@10: Ci[WS(csi, 2)] = FMA(KP866025403, To - (Tu + Tt), KP500000000 * (Tw - Tv)); Chris@10: Cr[WS(csr, 2)] = FMA(KP500000000, Tt - Tu, Tr) + (KP866025403 * (Tv + Tw)); Chris@10: } Chris@10: } Chris@10: } Chris@10: } Chris@10: Chris@10: static const kr2c_desc desc = { 9, "r2cfII_9", {25, 13, 17, 0}, &GENUS }; Chris@10: Chris@10: void X(codelet_r2cfII_9) (planner *p) { Chris@10: X(kr2c_register) (p, r2cfII_9, &desc); Chris@10: } Chris@10: Chris@10: #endif /* HAVE_FMA */