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 12 -name r2cb_12 -include r2cb.h */ Chris@19: Chris@19: /* Chris@19: * This function contains 38 FP additions, 16 FP multiplications, Chris@19: * (or, 22 additions, 0 multiplications, 16 fused multiply/add), Chris@19: * 31 stack variables, 2 constants, and 24 memory accesses Chris@19: */ Chris@19: #include "r2cb.h" Chris@19: Chris@19: static void r2cb_12(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_732050807, +1.732050807568877293527446341505872366942805254); Chris@19: DK(KP2_000000000, +2.000000000000000000000000000000000000000000000); 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(48, rs), MAKE_VOLATILE_STRIDE(48, csr), MAKE_VOLATILE_STRIDE(48, csi)) { Chris@19: E Ts, Tr; Chris@19: { Chris@19: E Tz, Te, Tn, Tk, Tc, Tw, Ty, Th, T4, T3, Td, T5; Chris@19: { Chris@19: E T8, Tu, Tl, Tm, Tb, T9, Ta, T1, T2, Tv; Chris@19: T8 = Cr[WS(csr, 3)]; Chris@19: T9 = Cr[WS(csr, 5)]; Chris@19: Ta = Cr[WS(csr, 1)]; Chris@19: Tu = Ci[WS(csi, 3)]; Chris@19: Tl = Ci[WS(csi, 5)]; Chris@19: Tm = Ci[WS(csi, 1)]; Chris@19: Tb = T9 + Ta; Chris@19: Tz = T9 - Ta; Chris@19: Te = Ci[WS(csi, 4)]; Chris@19: Tn = Tl - Tm; Chris@19: Tv = Tl + Tm; Chris@19: Tk = FNMS(KP2_000000000, T8, Tb); Chris@19: Tc = T8 + Tb; Chris@19: T1 = Cr[0]; Chris@19: T2 = Cr[WS(csr, 4)]; Chris@19: Tw = Tu - Tv; Chris@19: Ty = FMA(KP2_000000000, Tu, Tv); Chris@19: Th = Ci[WS(csi, 2)]; Chris@19: T4 = Cr[WS(csr, 6)]; Chris@19: T3 = FMA(KP2_000000000, T2, T1); Chris@19: Td = T1 - T2; Chris@19: T5 = Cr[WS(csr, 2)]; Chris@19: } Chris@19: { Chris@19: E To, Tp, Tf, Tg, T6, TA, TC; Chris@19: To = FMA(KP1_732050807, Tn, Tk); Chris@19: Ts = FNMS(KP1_732050807, Tn, Tk); Chris@19: Tp = FNMS(KP1_732050807, Te, Td); Chris@19: Tf = FMA(KP1_732050807, Te, Td); Chris@19: Tg = T4 - T5; Chris@19: T6 = FMA(KP2_000000000, T5, T4); Chris@19: TA = FMA(KP1_732050807, Tz, Ty); Chris@19: TC = FNMS(KP1_732050807, Tz, Ty); Chris@19: { Chris@19: E Tt, T7, Ti, Tq, Tj, TB, Tx; Chris@19: Tt = T3 - T6; Chris@19: T7 = T3 + T6; Chris@19: Ti = FNMS(KP1_732050807, Th, Tg); Chris@19: Tq = FMA(KP1_732050807, Th, Tg); Chris@19: R0[0] = FMA(KP2_000000000, Tc, T7); Chris@19: R0[WS(rs, 3)] = FNMS(KP2_000000000, Tc, T7); Chris@19: Tj = Tf + Ti; Chris@19: TB = Tf - Ti; Chris@19: Tr = Tp + Tq; Chris@19: Tx = Tp - Tq; Chris@19: R1[WS(rs, 5)] = TB + TC; Chris@19: R1[WS(rs, 2)] = TB - TC; Chris@19: R0[WS(rs, 4)] = Tj - To; Chris@19: R0[WS(rs, 1)] = Tj + To; Chris@19: R1[WS(rs, 3)] = Tx + TA; Chris@19: R1[0] = Tx - TA; Chris@19: R1[WS(rs, 4)] = FNMS(KP2_000000000, Tw, Tt); Chris@19: R1[WS(rs, 1)] = FMA(KP2_000000000, Tw, Tt); Chris@19: } Chris@19: } Chris@19: } Chris@19: R0[WS(rs, 2)] = Tr - Ts; Chris@19: R0[WS(rs, 5)] = Tr + Ts; Chris@19: } Chris@19: } Chris@19: } Chris@19: Chris@19: static const kr2c_desc desc = { 12, "r2cb_12", {22, 0, 16, 0}, &GENUS }; Chris@19: Chris@19: void X(codelet_r2cb_12) (planner *p) { Chris@19: X(kr2c_register) (p, r2cb_12, &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 12 -name r2cb_12 -include r2cb.h */ Chris@19: Chris@19: /* Chris@19: * This function contains 38 FP additions, 10 FP multiplications, Chris@19: * (or, 34 additions, 6 multiplications, 4 fused multiply/add), Chris@19: * 25 stack variables, 2 constants, and 24 memory accesses Chris@19: */ Chris@19: #include "r2cb.h" Chris@19: Chris@19: static void r2cb_12(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_732050807, +1.732050807568877293527446341505872366942805254); Chris@19: DK(KP2_000000000, +2.000000000000000000000000000000000000000000000); 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(48, rs), MAKE_VOLATILE_STRIDE(48, csr), MAKE_VOLATILE_STRIDE(48, csi)) { Chris@19: E T8, Tb, Tm, TA, Tw, Tx, Tp, TB, T3, Tr, Tg, T6, Ts, Tk; Chris@19: { Chris@19: E T9, Ta, Tn, To; Chris@19: T8 = Cr[WS(csr, 3)]; Chris@19: T9 = Cr[WS(csr, 5)]; Chris@19: Ta = Cr[WS(csr, 1)]; Chris@19: Tb = T9 + Ta; Chris@19: Tm = FMS(KP2_000000000, T8, Tb); Chris@19: TA = KP1_732050807 * (T9 - Ta); Chris@19: Tw = Ci[WS(csi, 3)]; Chris@19: Tn = Ci[WS(csi, 5)]; Chris@19: To = Ci[WS(csi, 1)]; Chris@19: Tx = Tn + To; Chris@19: Tp = KP1_732050807 * (Tn - To); Chris@19: TB = FMA(KP2_000000000, Tw, Tx); Chris@19: } Chris@19: { Chris@19: E Tf, T1, T2, Td, Te; Chris@19: Te = Ci[WS(csi, 4)]; Chris@19: Tf = KP1_732050807 * Te; Chris@19: T1 = Cr[0]; Chris@19: T2 = Cr[WS(csr, 4)]; Chris@19: Td = T1 - T2; Chris@19: T3 = FMA(KP2_000000000, T2, T1); Chris@19: Tr = Td - Tf; Chris@19: Tg = Td + Tf; Chris@19: } Chris@19: { Chris@19: E Tj, T4, T5, Th, Ti; Chris@19: Ti = Ci[WS(csi, 2)]; Chris@19: Tj = KP1_732050807 * Ti; Chris@19: T4 = Cr[WS(csr, 6)]; Chris@19: T5 = Cr[WS(csr, 2)]; Chris@19: Th = T4 - T5; Chris@19: T6 = FMA(KP2_000000000, T5, T4); Chris@19: Ts = Th + Tj; Chris@19: Tk = Th - Tj; Chris@19: } Chris@19: { Chris@19: E T7, Tc, Tz, TC; Chris@19: T7 = T3 + T6; Chris@19: Tc = KP2_000000000 * (T8 + Tb); Chris@19: R0[WS(rs, 3)] = T7 - Tc; Chris@19: R0[0] = T7 + Tc; Chris@19: { Chris@19: E Tl, Tq, TD, TE; Chris@19: Tl = Tg + Tk; Chris@19: Tq = Tm - Tp; Chris@19: R0[WS(rs, 1)] = Tl - Tq; Chris@19: R0[WS(rs, 4)] = Tl + Tq; Chris@19: TD = Tg - Tk; Chris@19: TE = TB - TA; Chris@19: R1[WS(rs, 2)] = TD - TE; Chris@19: R1[WS(rs, 5)] = TD + TE; Chris@19: } Chris@19: Tz = Tr - Ts; Chris@19: TC = TA + TB; Chris@19: R1[0] = Tz - TC; Chris@19: R1[WS(rs, 3)] = Tz + TC; Chris@19: { Chris@19: E Tv, Ty, Tt, Tu; Chris@19: Tv = T3 - T6; Chris@19: Ty = KP2_000000000 * (Tw - Tx); Chris@19: R1[WS(rs, 4)] = Tv - Ty; Chris@19: R1[WS(rs, 1)] = Tv + Ty; Chris@19: Tt = Tr + Ts; Chris@19: Tu = Tm + Tp; Chris@19: R0[WS(rs, 5)] = Tt - Tu; Chris@19: R0[WS(rs, 2)] = Tt + Tu; Chris@19: } Chris@19: } Chris@19: } Chris@19: } Chris@19: } Chris@19: Chris@19: static const kr2c_desc desc = { 12, "r2cb_12", {34, 6, 4, 0}, &GENUS }; Chris@19: Chris@19: void X(codelet_r2cb_12) (planner *p) { Chris@19: X(kr2c_register) (p, r2cb_12, &desc); Chris@19: } Chris@19: Chris@19: #endif /* HAVE_FMA */