Chris@19: /* Chris@19: * Copyright (c) 2003, 2007-14 Matteo Frigo Chris@19: * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology Chris@19: * Chris@19: * This program is free software; you can redistribute it and/or modify Chris@19: * it under the terms of the GNU General Public License as published by Chris@19: * the Free Software Foundation; either version 2 of the License, or Chris@19: * (at your option) any later version. Chris@19: * Chris@19: * This program is distributed in the hope that it will be useful, Chris@19: * but WITHOUT ANY WARRANTY; without even the implied warranty of Chris@19: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the Chris@19: * GNU General Public License for more details. Chris@19: * Chris@19: * You should have received a copy of the GNU General Public License Chris@19: * along with this program; if not, write to the Free Software Chris@19: * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA Chris@19: * Chris@19: */ Chris@19: Chris@19: /* This file was automatically generated --- DO NOT EDIT */ Chris@19: /* Generated on Tue Mar 4 13:50:24 EST 2014 */ Chris@19: Chris@19: #include "codelet-rdft.h" Chris@19: Chris@19: #ifdef HAVE_FMA Chris@19: Chris@19: /* Generated by: ../../../genfft/gen_r2cb.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -sign 1 -n 10 -name r2cb_10 -include r2cb.h */ Chris@19: Chris@19: /* Chris@19: * This function contains 34 FP additions, 20 FP multiplications, Chris@19: * (or, 14 additions, 0 multiplications, 20 fused multiply/add), Chris@19: * 30 stack variables, 5 constants, and 20 memory accesses Chris@19: */ Chris@19: #include "r2cb.h" Chris@19: Chris@19: static void r2cb_10(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs) Chris@19: { Chris@19: DK(KP1_902113032, +1.902113032590307144232878666758764286811397268); Chris@19: DK(KP1_118033988, +1.118033988749894848204586834365638117720309180); Chris@19: DK(KP2_000000000, +2.000000000000000000000000000000000000000000000); Chris@19: DK(KP500000000, +0.500000000000000000000000000000000000000000000); Chris@19: DK(KP618033988, +0.618033988749894848204586834365638117720309180); Chris@19: { Chris@19: INT i; Chris@19: for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(40, rs), MAKE_VOLATILE_STRIDE(40, csr), MAKE_VOLATILE_STRIDE(40, csi)) { Chris@19: E Tb, T3, Tc, T6, Tq, To, Ty, Tw, Td, T9; Chris@19: { Chris@19: E Tu, Tn, T7, Tv, Tk, T8; Chris@19: { Chris@19: E T1, T2, Tl, Tm; Chris@19: T1 = Cr[0]; Chris@19: T2 = Cr[WS(csr, 5)]; Chris@19: Tl = Ci[WS(csi, 2)]; Chris@19: Tm = Ci[WS(csi, 3)]; Chris@19: { Chris@19: E Ti, Tj, T4, T5; Chris@19: Ti = Ci[WS(csi, 4)]; Chris@19: Tb = T1 + T2; Chris@19: T3 = T1 - T2; Chris@19: Tu = Tl + Tm; Chris@19: Tn = Tl - Tm; Chris@19: Tj = Ci[WS(csi, 1)]; Chris@19: T4 = Cr[WS(csr, 2)]; Chris@19: T5 = Cr[WS(csr, 3)]; Chris@19: T7 = Cr[WS(csr, 4)]; Chris@19: Tv = Ti + Tj; Chris@19: Tk = Ti - Tj; Chris@19: Tc = T4 + T5; Chris@19: T6 = T4 - T5; Chris@19: T8 = Cr[WS(csr, 1)]; Chris@19: } Chris@19: } Chris@19: Tq = FMA(KP618033988, Tk, Tn); Chris@19: To = FNMS(KP618033988, Tn, Tk); Chris@19: Ty = FNMS(KP618033988, Tu, Tv); Chris@19: Tw = FMA(KP618033988, Tv, Tu); Chris@19: Td = T7 + T8; Chris@19: T9 = T7 - T8; Chris@19: } Chris@19: { Chris@19: E Te, Tg, Ta, Ts, Tf, Tr; Chris@19: Te = Tc + Td; Chris@19: Tg = Tc - Td; Chris@19: Ta = T6 + T9; Chris@19: Ts = T6 - T9; Chris@19: Tf = FNMS(KP500000000, Te, Tb); Chris@19: R0[0] = FMA(KP2_000000000, Te, Tb); Chris@19: Tr = FNMS(KP500000000, Ta, T3); Chris@19: R1[WS(rs, 2)] = FMA(KP2_000000000, Ta, T3); Chris@19: { Chris@19: E Th, Tp, Tt, Tx; Chris@19: Th = FNMS(KP1_118033988, Tg, Tf); Chris@19: Tp = FMA(KP1_118033988, Tg, Tf); Chris@19: Tt = FMA(KP1_118033988, Ts, Tr); Chris@19: Tx = FNMS(KP1_118033988, Ts, Tr); Chris@19: R0[WS(rs, 3)] = FNMS(KP1_902113032, Tq, Tp); Chris@19: R0[WS(rs, 2)] = FMA(KP1_902113032, Tq, Tp); Chris@19: R0[WS(rs, 1)] = FMA(KP1_902113032, To, Th); Chris@19: R0[WS(rs, 4)] = FNMS(KP1_902113032, To, Th); Chris@19: R1[WS(rs, 1)] = FNMS(KP1_902113032, Ty, Tx); Chris@19: R1[WS(rs, 3)] = FMA(KP1_902113032, Ty, Tx); Chris@19: R1[WS(rs, 4)] = FMA(KP1_902113032, Tw, Tt); Chris@19: R1[0] = FNMS(KP1_902113032, Tw, Tt); Chris@19: } Chris@19: } Chris@19: } Chris@19: } Chris@19: } Chris@19: Chris@19: static const kr2c_desc desc = { 10, "r2cb_10", {14, 0, 20, 0}, &GENUS }; Chris@19: Chris@19: void X(codelet_r2cb_10) (planner *p) { Chris@19: X(kr2c_register) (p, r2cb_10, &desc); Chris@19: } Chris@19: Chris@19: #else /* HAVE_FMA */ Chris@19: Chris@19: /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 10 -name r2cb_10 -include r2cb.h */ Chris@19: Chris@19: /* Chris@19: * This function contains 34 FP additions, 14 FP multiplications, Chris@19: * (or, 26 additions, 6 multiplications, 8 fused multiply/add), Chris@19: * 26 stack variables, 5 constants, and 20 memory accesses Chris@19: */ Chris@19: #include "r2cb.h" Chris@19: Chris@19: static void r2cb_10(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs) Chris@19: { Chris@19: DK(KP500000000, +0.500000000000000000000000000000000000000000000); Chris@19: DK(KP1_902113032, +1.902113032590307144232878666758764286811397268); Chris@19: DK(KP1_175570504, +1.175570504584946258337411909278145537195304875); Chris@19: DK(KP2_000000000, +2.000000000000000000000000000000000000000000000); Chris@19: DK(KP1_118033988, +1.118033988749894848204586834365638117720309180); Chris@19: { Chris@19: INT i; Chris@19: for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(40, rs), MAKE_VOLATILE_STRIDE(40, csr), MAKE_VOLATILE_STRIDE(40, csi)) { Chris@19: E T3, Tb, Tn, Tv, Tk, Tu, Ta, Ts, Te, Tg, Ti, Tj; Chris@19: { Chris@19: E T1, T2, Tl, Tm; Chris@19: T1 = Cr[0]; Chris@19: T2 = Cr[WS(csr, 5)]; Chris@19: T3 = T1 - T2; Chris@19: Tb = T1 + T2; Chris@19: Tl = Ci[WS(csi, 4)]; Chris@19: Tm = Ci[WS(csi, 1)]; Chris@19: Tn = Tl - Tm; Chris@19: Tv = Tl + Tm; Chris@19: } Chris@19: Ti = Ci[WS(csi, 2)]; Chris@19: Tj = Ci[WS(csi, 3)]; Chris@19: Tk = Ti - Tj; Chris@19: Tu = Ti + Tj; Chris@19: { Chris@19: E T6, Tc, T9, Td; Chris@19: { Chris@19: E T4, T5, T7, T8; Chris@19: T4 = Cr[WS(csr, 2)]; Chris@19: T5 = Cr[WS(csr, 3)]; Chris@19: T6 = T4 - T5; Chris@19: Tc = T4 + T5; Chris@19: T7 = Cr[WS(csr, 4)]; Chris@19: T8 = Cr[WS(csr, 1)]; Chris@19: T9 = T7 - T8; Chris@19: Td = T7 + T8; Chris@19: } Chris@19: Ta = T6 + T9; Chris@19: Ts = KP1_118033988 * (T6 - T9); Chris@19: Te = Tc + Td; Chris@19: Tg = KP1_118033988 * (Tc - Td); Chris@19: } Chris@19: R1[WS(rs, 2)] = FMA(KP2_000000000, Ta, T3); Chris@19: R0[0] = FMA(KP2_000000000, Te, Tb); Chris@19: { Chris@19: E To, Tq, Th, Tp, Tf; Chris@19: To = FNMS(KP1_902113032, Tn, KP1_175570504 * Tk); Chris@19: Tq = FMA(KP1_902113032, Tk, KP1_175570504 * Tn); Chris@19: Tf = FNMS(KP500000000, Te, Tb); Chris@19: Th = Tf - Tg; Chris@19: Tp = Tg + Tf; Chris@19: R0[WS(rs, 1)] = Th - To; Chris@19: R0[WS(rs, 2)] = Tp + Tq; Chris@19: R0[WS(rs, 4)] = Th + To; Chris@19: R0[WS(rs, 3)] = Tp - Tq; Chris@19: } Chris@19: { Chris@19: E Tw, Ty, Tt, Tx, Tr; Chris@19: Tw = FNMS(KP1_902113032, Tv, KP1_175570504 * Tu); Chris@19: Ty = FMA(KP1_902113032, Tu, KP1_175570504 * Tv); Chris@19: Tr = FNMS(KP500000000, Ta, T3); Chris@19: Tt = Tr - Ts; Chris@19: Tx = Ts + Tr; Chris@19: R1[WS(rs, 3)] = Tt - Tw; Chris@19: R1[WS(rs, 4)] = Tx + Ty; Chris@19: R1[WS(rs, 1)] = Tt + Tw; Chris@19: R1[0] = Tx - Ty; Chris@19: } Chris@19: } Chris@19: } Chris@19: } Chris@19: Chris@19: static const kr2c_desc desc = { 10, "r2cb_10", {26, 6, 8, 0}, &GENUS }; Chris@19: Chris@19: void X(codelet_r2cb_10) (planner *p) { Chris@19: X(kr2c_register) (p, r2cb_10, &desc); Chris@19: } Chris@19: Chris@19: #endif /* HAVE_FMA */