/* * Copyright (c) 2003, 2007-8 Matteo Frigo * Copyright (c) 2003, 2007-8 Massachusetts Institute of Technology * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 2 of the License, or * (at your option) any later version. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA * */ /* This file was automatically generated --- DO NOT EDIT */ /* Generated on Sun Jul 12 06:42:47 EDT 2009 */ #include "codelet-dft.h" #ifdef HAVE_FMA /* Generated by: ../../../genfft/gen_twiddle_c -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 8 -name t2bv_8 -include t2b.h -sign 1 */ /* * This function contains 33 FP additions, 24 FP multiplications, * (or, 23 additions, 14 multiplications, 10 fused multiply/add), * 36 stack variables, 1 constants, and 16 memory accesses */ #include "t2b.h" static void t2bv_8(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP707106781, +0.707106781186547524400844362104849039284835938); INT m; R *x; x = ii; for (m = mb, W = W + (mb * ((TWVL / VL) * 14)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 14), MAKE_VOLATILE_STRIDE(rs)) { V T1, T2, Th, Tj, T5, T7, Ta, Tc; T1 = LD(&(x[0]), ms, &(x[0])); T2 = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Th = LD(&(x[WS(rs, 2)]), ms, &(x[0])); Tj = LD(&(x[WS(rs, 6)]), ms, &(x[0])); T5 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); T7 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); Ta = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); Tc = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); { V T3, Ti, Tk, T6, T8, Tb, Td; T3 = BYTW(&(W[TWVL * 6]), T2); Ti = BYTW(&(W[TWVL * 2]), Th); Tk = BYTW(&(W[TWVL * 10]), Tj); T6 = BYTW(&(W[0]), T5); T8 = BYTW(&(W[TWVL * 8]), T7); Tb = BYTW(&(W[TWVL * 12]), Ta); Td = BYTW(&(W[TWVL * 4]), Tc); { V Tq, T4, Tr, Tl, Tt, T9, Tu, Te, Tw, Ts; Tq = VADD(T1, T3); T4 = VSUB(T1, T3); Tr = VADD(Ti, Tk); Tl = VSUB(Ti, Tk); Tt = VADD(T6, T8); T9 = VSUB(T6, T8); Tu = VADD(Tb, Td); Te = VSUB(Tb, Td); Tw = VADD(Tq, Tr); Ts = VSUB(Tq, Tr); { V Tx, Tv, Tm, Tf; Tx = VADD(Tt, Tu); Tv = VSUB(Tt, Tu); Tm = VSUB(T9, Te); Tf = VADD(T9, Te); { V Tp, Tn, To, Tg; ST(&(x[0]), VADD(Tw, Tx), ms, &(x[0])); ST(&(x[WS(rs, 4)]), VSUB(Tw, Tx), ms, &(x[0])); ST(&(x[WS(rs, 2)]), VFMAI(Tv, Ts), ms, &(x[0])); ST(&(x[WS(rs, 6)]), VFNMSI(Tv, Ts), ms, &(x[0])); Tp = VFMA(LDK(KP707106781), Tm, Tl); Tn = VFNMS(LDK(KP707106781), Tm, Tl); To = VFMA(LDK(KP707106781), Tf, T4); Tg = VFNMS(LDK(KP707106781), Tf, T4); ST(&(x[WS(rs, 1)]), VFMAI(Tp, To), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 7)]), VFNMSI(Tp, To), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 5)]), VFMAI(Tn, Tg), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 3)]), VFNMSI(Tn, Tg), ms, &(x[WS(rs, 1)])); } } } } } } static const tw_instr twinstr[] = { VTW(0, 1), VTW(0, 2), VTW(0, 3), VTW(0, 4), VTW(0, 5), VTW(0, 6), VTW(0, 7), {TW_NEXT, VL, 0} }; static const ct_desc desc = { 8, "t2bv_8", twinstr, &GENUS, {23, 14, 10, 0}, 0, 0, 0 }; void X(codelet_t2bv_8) (planner *p) { X(kdft_dit_register) (p, t2bv_8, &desc); } #else /* HAVE_FMA */ /* Generated by: ../../../genfft/gen_twiddle_c -simd -compact -variables 4 -pipeline-latency 8 -n 8 -name t2bv_8 -include t2b.h -sign 1 */ /* * This function contains 33 FP additions, 16 FP multiplications, * (or, 33 additions, 16 multiplications, 0 fused multiply/add), * 24 stack variables, 1 constants, and 16 memory accesses */ #include "t2b.h" static void t2bv_8(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP707106781, +0.707106781186547524400844362104849039284835938); INT m; R *x; x = ii; for (m = mb, W = W + (mb * ((TWVL / VL) * 14)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 14), MAKE_VOLATILE_STRIDE(rs)) { V Tl, Tq, Tg, Tr, T5, Tt, Ta, Tu, Ti, Tk, Tj; Ti = LD(&(x[0]), ms, &(x[0])); Tj = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Tk = BYTW(&(W[TWVL * 6]), Tj); Tl = VSUB(Ti, Tk); Tq = VADD(Ti, Tk); { V Td, Tf, Tc, Te; Tc = LD(&(x[WS(rs, 2)]), ms, &(x[0])); Td = BYTW(&(W[TWVL * 2]), Tc); Te = LD(&(x[WS(rs, 6)]), ms, &(x[0])); Tf = BYTW(&(W[TWVL * 10]), Te); Tg = VSUB(Td, Tf); Tr = VADD(Td, Tf); } { V T2, T4, T1, T3; T1 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); T2 = BYTW(&(W[0]), T1); T3 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); T4 = BYTW(&(W[TWVL * 8]), T3); T5 = VSUB(T2, T4); Tt = VADD(T2, T4); } { V T7, T9, T6, T8; T6 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); T7 = BYTW(&(W[TWVL * 12]), T6); T8 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); T9 = BYTW(&(W[TWVL * 4]), T8); Ta = VSUB(T7, T9); Tu = VADD(T7, T9); } { V Ts, Tv, Tw, Tx; Ts = VSUB(Tq, Tr); Tv = VBYI(VSUB(Tt, Tu)); ST(&(x[WS(rs, 6)]), VSUB(Ts, Tv), ms, &(x[0])); ST(&(x[WS(rs, 2)]), VADD(Ts, Tv), ms, &(x[0])); Tw = VADD(Tq, Tr); Tx = VADD(Tt, Tu); ST(&(x[WS(rs, 4)]), VSUB(Tw, Tx), ms, &(x[0])); ST(&(x[0]), VADD(Tw, Tx), ms, &(x[0])); { V Th, To, Tn, Tp, Tb, Tm; Tb = VMUL(LDK(KP707106781), VSUB(T5, Ta)); Th = VBYI(VSUB(Tb, Tg)); To = VBYI(VADD(Tg, Tb)); Tm = VMUL(LDK(KP707106781), VADD(T5, Ta)); Tn = VSUB(Tl, Tm); Tp = VADD(Tl, Tm); ST(&(x[WS(rs, 3)]), VADD(Th, Tn), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 7)]), VSUB(Tp, To), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 5)]), VSUB(Tn, Th), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 1)]), VADD(To, Tp), ms, &(x[WS(rs, 1)])); } } } } static const tw_instr twinstr[] = { VTW(0, 1), VTW(0, 2), VTW(0, 3), VTW(0, 4), VTW(0, 5), VTW(0, 6), VTW(0, 7), {TW_NEXT, VL, 0} }; static const ct_desc desc = { 8, "t2bv_8", twinstr, &GENUS, {33, 16, 0, 0}, 0, 0, 0 }; void X(codelet_t2bv_8) (planner *p) { X(kdft_dit_register) (p, t2bv_8, &desc); } #endif /* HAVE_FMA */