cannam@95: /* cannam@95: * Copyright (c) 2003, 2007-11 Matteo Frigo cannam@95: * Copyright (c) 2003, 2007-11 Massachusetts Institute of Technology cannam@95: * cannam@95: * This program is free software; you can redistribute it and/or modify cannam@95: * it under the terms of the GNU General Public License as published by cannam@95: * the Free Software Foundation; either version 2 of the License, or cannam@95: * (at your option) any later version. cannam@95: * cannam@95: * This program is distributed in the hope that it will be useful, cannam@95: * but WITHOUT ANY WARRANTY; without even the implied warranty of cannam@95: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the cannam@95: * GNU General Public License for more details. cannam@95: * cannam@95: * You should have received a copy of the GNU General Public License cannam@95: * along with this program; if not, write to the Free Software cannam@95: * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA cannam@95: * cannam@95: */ cannam@95: cannam@95: /* This file was automatically generated --- DO NOT EDIT */ cannam@95: /* Generated on Sun Nov 25 07:40:14 EST 2012 */ cannam@95: cannam@95: #include "codelet-rdft.h" cannam@95: cannam@95: #ifdef HAVE_FMA cannam@95: cannam@95: /* Generated by: ../../../genfft/gen_r2cf.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 8 -name r2cfII_8 -dft-II -include r2cfII.h */ cannam@95: cannam@95: /* cannam@95: * This function contains 22 FP additions, 16 FP multiplications, cannam@95: * (or, 6 additions, 0 multiplications, 16 fused multiply/add), cannam@95: * 22 stack variables, 3 constants, and 16 memory accesses cannam@95: */ cannam@95: #include "r2cfII.h" cannam@95: cannam@95: static void r2cfII_8(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs) cannam@95: { cannam@95: DK(KP923879532, +0.923879532511286756128183189396788286822416626); cannam@95: DK(KP707106781, +0.707106781186547524400844362104849039284835938); cannam@95: DK(KP414213562, +0.414213562373095048801688724209698078569671875); cannam@95: { cannam@95: INT i; cannam@95: for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(32, rs), MAKE_VOLATILE_STRIDE(32, csr), MAKE_VOLATILE_STRIDE(32, csi)) { cannam@95: E Te, T8, Td, T5, Tj, Tl, Tf, Tb; cannam@95: { cannam@95: E T1, Th, T9, Ti, T4, Ta; cannam@95: T1 = R0[0]; cannam@95: Th = R0[WS(rs, 2)]; cannam@95: { cannam@95: E T2, T3, T6, T7; cannam@95: T2 = R0[WS(rs, 1)]; cannam@95: T3 = R0[WS(rs, 3)]; cannam@95: T6 = R1[0]; cannam@95: T7 = R1[WS(rs, 2)]; cannam@95: T9 = R1[WS(rs, 3)]; cannam@95: Ti = T2 + T3; cannam@95: T4 = T2 - T3; cannam@95: Te = FMA(KP414213562, T6, T7); cannam@95: T8 = FNMS(KP414213562, T7, T6); cannam@95: Ta = R1[WS(rs, 1)]; cannam@95: } cannam@95: Td = FNMS(KP707106781, T4, T1); cannam@95: T5 = FMA(KP707106781, T4, T1); cannam@95: Tj = FMA(KP707106781, Ti, Th); cannam@95: Tl = FNMS(KP707106781, Ti, Th); cannam@95: Tf = FMA(KP414213562, T9, Ta); cannam@95: Tb = FMS(KP414213562, Ta, T9); cannam@95: } cannam@95: { cannam@95: E Tk, Tg, Tc, Tm; cannam@95: Tk = Te + Tf; cannam@95: Tg = Te - Tf; cannam@95: Tc = T8 + Tb; cannam@95: Tm = Tb - T8; cannam@95: Cr[WS(csr, 1)] = FMA(KP923879532, Tg, Td); cannam@95: Cr[WS(csr, 2)] = FNMS(KP923879532, Tg, Td); cannam@95: Ci[WS(csi, 3)] = FNMS(KP923879532, Tk, Tj); cannam@95: Ci[0] = -(FMA(KP923879532, Tk, Tj)); cannam@95: Ci[WS(csi, 1)] = FMA(KP923879532, Tm, Tl); cannam@95: Ci[WS(csi, 2)] = FMS(KP923879532, Tm, Tl); cannam@95: Cr[0] = FMA(KP923879532, Tc, T5); cannam@95: Cr[WS(csr, 3)] = FNMS(KP923879532, Tc, T5); cannam@95: } cannam@95: } cannam@95: } cannam@95: } cannam@95: cannam@95: static const kr2c_desc desc = { 8, "r2cfII_8", {6, 0, 16, 0}, &GENUS }; cannam@95: cannam@95: void X(codelet_r2cfII_8) (planner *p) { cannam@95: X(kr2c_register) (p, r2cfII_8, &desc); cannam@95: } cannam@95: cannam@95: #else /* HAVE_FMA */ cannam@95: cannam@95: /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 8 -name r2cfII_8 -dft-II -include r2cfII.h */ cannam@95: cannam@95: /* cannam@95: * This function contains 22 FP additions, 10 FP multiplications, cannam@95: * (or, 18 additions, 6 multiplications, 4 fused multiply/add), cannam@95: * 18 stack variables, 3 constants, and 16 memory accesses cannam@95: */ cannam@95: #include "r2cfII.h" cannam@95: cannam@95: static void r2cfII_8(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs) cannam@95: { cannam@95: DK(KP382683432, +0.382683432365089771728459984030398866761344562); cannam@95: DK(KP923879532, +0.923879532511286756128183189396788286822416626); cannam@95: DK(KP707106781, +0.707106781186547524400844362104849039284835938); cannam@95: { cannam@95: INT i; cannam@95: for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(32, rs), MAKE_VOLATILE_STRIDE(32, csr), MAKE_VOLATILE_STRIDE(32, csi)) { cannam@95: E T1, Tj, T4, Ti, T8, Te, Tb, Tf, T2, T3; cannam@95: T1 = R0[0]; cannam@95: Tj = R0[WS(rs, 2)]; cannam@95: T2 = R0[WS(rs, 1)]; cannam@95: T3 = R0[WS(rs, 3)]; cannam@95: T4 = KP707106781 * (T2 - T3); cannam@95: Ti = KP707106781 * (T2 + T3); cannam@95: { cannam@95: E T6, T7, T9, Ta; cannam@95: T6 = R1[0]; cannam@95: T7 = R1[WS(rs, 2)]; cannam@95: T8 = FNMS(KP382683432, T7, KP923879532 * T6); cannam@95: Te = FMA(KP382683432, T6, KP923879532 * T7); cannam@95: T9 = R1[WS(rs, 1)]; cannam@95: Ta = R1[WS(rs, 3)]; cannam@95: Tb = FNMS(KP923879532, Ta, KP382683432 * T9); cannam@95: Tf = FMA(KP923879532, T9, KP382683432 * Ta); cannam@95: } cannam@95: { cannam@95: E T5, Tc, Th, Tk; cannam@95: T5 = T1 + T4; cannam@95: Tc = T8 + Tb; cannam@95: Cr[WS(csr, 3)] = T5 - Tc; cannam@95: Cr[0] = T5 + Tc; cannam@95: Th = Te + Tf; cannam@95: Tk = Ti + Tj; cannam@95: Ci[0] = -(Th + Tk); cannam@95: Ci[WS(csi, 3)] = Tk - Th; cannam@95: } cannam@95: { cannam@95: E Td, Tg, Tl, Tm; cannam@95: Td = T1 - T4; cannam@95: Tg = Te - Tf; cannam@95: Cr[WS(csr, 2)] = Td - Tg; cannam@95: Cr[WS(csr, 1)] = Td + Tg; cannam@95: Tl = Tb - T8; cannam@95: Tm = Tj - Ti; cannam@95: Ci[WS(csi, 2)] = Tl - Tm; cannam@95: Ci[WS(csi, 1)] = Tl + Tm; cannam@95: } cannam@95: } cannam@95: } cannam@95: } cannam@95: cannam@95: static const kr2c_desc desc = { 8, "r2cfII_8", {18, 6, 4, 0}, &GENUS }; cannam@95: cannam@95: void X(codelet_r2cfII_8) (planner *p) { cannam@95: X(kr2c_register) (p, r2cfII_8, &desc); cannam@95: } cannam@95: cannam@95: #endif /* HAVE_FMA */