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 13:45:51 EST 2014 */ Chris@19: Chris@19: #include "codelet-dft.h" Chris@19: Chris@19: #ifdef HAVE_FMA Chris@19: Chris@19: /* Generated by: ../../../genfft/gen_twiddle.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 4 -name t1_4 -include t.h */ Chris@19: Chris@19: /* Chris@19: * This function contains 22 FP additions, 12 FP multiplications, Chris@19: * (or, 16 additions, 6 multiplications, 6 fused multiply/add), Chris@19: * 31 stack variables, 0 constants, and 16 memory accesses Chris@19: */ Chris@19: #include "t.h" Chris@19: Chris@19: static void t1_4(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) Chris@19: { Chris@19: { Chris@19: INT m; Chris@19: for (m = mb, W = W + (mb * 6); m < me; m = m + 1, ri = ri + ms, ii = ii + ms, W = W + 6, MAKE_VOLATILE_STRIDE(8, rs)) { Chris@19: E To, Te, Tm, T8, Tw, Tx, Tq, Tk; Chris@19: { Chris@19: E T1, Tv, Tu, T7, Tg, Tj, Tf, Ti, Tp, Th; Chris@19: T1 = ri[0]; Chris@19: Tv = ii[0]; Chris@19: { Chris@19: E T3, T6, T2, T5; Chris@19: T3 = ri[WS(rs, 2)]; Chris@19: T6 = ii[WS(rs, 2)]; Chris@19: T2 = W[2]; Chris@19: T5 = W[3]; Chris@19: { Chris@19: E Ta, Td, Tc, Tn, Tb, Tt, T4, T9; Chris@19: Ta = ri[WS(rs, 1)]; Chris@19: Td = ii[WS(rs, 1)]; Chris@19: Tt = T2 * T6; Chris@19: T4 = T2 * T3; Chris@19: T9 = W[0]; Chris@19: Tc = W[1]; Chris@19: Tu = FNMS(T5, T3, Tt); Chris@19: T7 = FMA(T5, T6, T4); Chris@19: Tn = T9 * Td; Chris@19: Tb = T9 * Ta; Chris@19: Tg = ri[WS(rs, 3)]; Chris@19: Tj = ii[WS(rs, 3)]; Chris@19: To = FNMS(Tc, Ta, Tn); Chris@19: Te = FMA(Tc, Td, Tb); Chris@19: Tf = W[4]; Chris@19: Ti = W[5]; Chris@19: } Chris@19: } Chris@19: Tm = T1 - T7; Chris@19: T8 = T1 + T7; Chris@19: Tw = Tu + Tv; Chris@19: Tx = Tv - Tu; Chris@19: Tp = Tf * Tj; Chris@19: Th = Tf * Tg; Chris@19: Tq = FNMS(Ti, Tg, Tp); Chris@19: Tk = FMA(Ti, Tj, Th); Chris@19: } Chris@19: { Chris@19: E Ts, Tr, Tl, Ty; Chris@19: Ts = To + Tq; Chris@19: Tr = To - Tq; Chris@19: Tl = Te + Tk; Chris@19: Ty = Te - Tk; Chris@19: ri[WS(rs, 1)] = Tm + Tr; Chris@19: ri[WS(rs, 3)] = Tm - Tr; Chris@19: ii[WS(rs, 2)] = Tw - Ts; Chris@19: ii[0] = Ts + Tw; Chris@19: ii[WS(rs, 3)] = Ty + Tx; Chris@19: ii[WS(rs, 1)] = Tx - Ty; Chris@19: ri[0] = T8 + Tl; Chris@19: ri[WS(rs, 2)] = T8 - Tl; Chris@19: } Chris@19: } Chris@19: } Chris@19: } Chris@19: Chris@19: static const tw_instr twinstr[] = { Chris@19: {TW_FULL, 0, 4}, Chris@19: {TW_NEXT, 1, 0} Chris@19: }; Chris@19: Chris@19: static const ct_desc desc = { 4, "t1_4", twinstr, &GENUS, {16, 6, 6, 0}, 0, 0, 0 }; Chris@19: Chris@19: void X(codelet_t1_4) (planner *p) { Chris@19: X(kdft_dit_register) (p, t1_4, &desc); Chris@19: } Chris@19: #else /* HAVE_FMA */ Chris@19: Chris@19: /* Generated by: ../../../genfft/gen_twiddle.native -compact -variables 4 -pipeline-latency 4 -n 4 -name t1_4 -include t.h */ Chris@19: Chris@19: /* Chris@19: * This function contains 22 FP additions, 12 FP multiplications, Chris@19: * (or, 16 additions, 6 multiplications, 6 fused multiply/add), Chris@19: * 13 stack variables, 0 constants, and 16 memory accesses Chris@19: */ Chris@19: #include "t.h" Chris@19: Chris@19: static void t1_4(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) Chris@19: { Chris@19: { Chris@19: INT m; Chris@19: for (m = mb, W = W + (mb * 6); m < me; m = m + 1, ri = ri + ms, ii = ii + ms, W = W + 6, MAKE_VOLATILE_STRIDE(8, rs)) { Chris@19: E T1, Tp, T6, To, Tc, Tk, Th, Tl; Chris@19: T1 = ri[0]; Chris@19: Tp = ii[0]; Chris@19: { Chris@19: E T3, T5, T2, T4; Chris@19: T3 = ri[WS(rs, 2)]; Chris@19: T5 = ii[WS(rs, 2)]; Chris@19: T2 = W[2]; Chris@19: T4 = W[3]; Chris@19: T6 = FMA(T2, T3, T4 * T5); Chris@19: To = FNMS(T4, T3, T2 * T5); Chris@19: } Chris@19: { Chris@19: E T9, Tb, T8, Ta; Chris@19: T9 = ri[WS(rs, 1)]; Chris@19: Tb = ii[WS(rs, 1)]; Chris@19: T8 = W[0]; Chris@19: Ta = W[1]; Chris@19: Tc = FMA(T8, T9, Ta * Tb); Chris@19: Tk = FNMS(Ta, T9, T8 * Tb); Chris@19: } Chris@19: { Chris@19: E Te, Tg, Td, Tf; Chris@19: Te = ri[WS(rs, 3)]; Chris@19: Tg = ii[WS(rs, 3)]; Chris@19: Td = W[4]; Chris@19: Tf = W[5]; Chris@19: Th = FMA(Td, Te, Tf * Tg); Chris@19: Tl = FNMS(Tf, Te, Td * Tg); Chris@19: } Chris@19: { Chris@19: E T7, Ti, Tn, Tq; Chris@19: T7 = T1 + T6; Chris@19: Ti = Tc + Th; Chris@19: ri[WS(rs, 2)] = T7 - Ti; Chris@19: ri[0] = T7 + Ti; Chris@19: Tn = Tk + Tl; Chris@19: Tq = To + Tp; Chris@19: ii[0] = Tn + Tq; Chris@19: ii[WS(rs, 2)] = Tq - Tn; Chris@19: } Chris@19: { Chris@19: E Tj, Tm, Tr, Ts; Chris@19: Tj = T1 - T6; Chris@19: Tm = Tk - Tl; Chris@19: ri[WS(rs, 3)] = Tj - Tm; Chris@19: ri[WS(rs, 1)] = Tj + Tm; Chris@19: Tr = Tp - To; Chris@19: Ts = Tc - Th; Chris@19: ii[WS(rs, 1)] = Tr - Ts; Chris@19: ii[WS(rs, 3)] = Ts + Tr; Chris@19: } Chris@19: } Chris@19: } Chris@19: } Chris@19: Chris@19: static const tw_instr twinstr[] = { Chris@19: {TW_FULL, 0, 4}, Chris@19: {TW_NEXT, 1, 0} Chris@19: }; Chris@19: Chris@19: static const ct_desc desc = { 4, "t1_4", twinstr, &GENUS, {16, 6, 6, 0}, 0, 0, 0 }; Chris@19: Chris@19: void X(codelet_t1_4) (planner *p) { Chris@19: X(kdft_dit_register) (p, t1_4, &desc); Chris@19: } Chris@19: #endif /* HAVE_FMA */