cannam@167: /* cannam@167: * Copyright (c) 2003, 2007-14 Matteo Frigo cannam@167: * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology cannam@167: * cannam@167: * This program is free software; you can redistribute it and/or modify cannam@167: * it under the terms of the GNU General Public License as published by cannam@167: * the Free Software Foundation; either version 2 of the License, or cannam@167: * (at your option) any later version. cannam@167: * cannam@167: * This program is distributed in the hope that it will be useful, cannam@167: * but WITHOUT ANY WARRANTY; without even the implied warranty of cannam@167: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the cannam@167: * GNU General Public License for more details. cannam@167: * cannam@167: * You should have received a copy of the GNU General Public License cannam@167: * along with this program; if not, write to the Free Software cannam@167: * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA cannam@167: * cannam@167: */ cannam@167: cannam@167: /* This file was automatically generated --- DO NOT EDIT */ cannam@167: /* Generated on Thu May 24 08:05:26 EDT 2018 */ cannam@167: cannam@167: #include "dft/codelet-dft.h" cannam@167: cannam@167: #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA) cannam@167: cannam@167: /* Generated by: ../../../genfft/gen_twiddle_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -n 6 -name t1fv_6 -include dft/simd/t1f.h */ cannam@167: cannam@167: /* cannam@167: * This function contains 23 FP additions, 18 FP multiplications, cannam@167: * (or, 17 additions, 12 multiplications, 6 fused multiply/add), cannam@167: * 19 stack variables, 2 constants, and 12 memory accesses cannam@167: */ cannam@167: #include "dft/simd/t1f.h" cannam@167: cannam@167: static void t1fv_6(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) cannam@167: { cannam@167: DVK(KP500000000, +0.500000000000000000000000000000000000000000000); cannam@167: DVK(KP866025403, +0.866025403784438646763723170752936183471402627); cannam@167: { cannam@167: INT m; cannam@167: R *x; cannam@167: x = ri; cannam@167: for (m = mb, W = W + (mb * ((TWVL / VL) * 10)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 10), MAKE_VOLATILE_STRIDE(6, rs)) { cannam@167: V T4, Ti, Te, Tk, T9, Tj, T1, T3, T2; cannam@167: T1 = LD(&(x[0]), ms, &(x[0])); cannam@167: T2 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); cannam@167: T3 = BYTWJ(&(W[TWVL * 4]), T2); cannam@167: T4 = VSUB(T1, T3); cannam@167: Ti = VADD(T1, T3); cannam@167: { cannam@167: V Tb, Td, Ta, Tc; cannam@167: Ta = LD(&(x[WS(rs, 4)]), ms, &(x[0])); cannam@167: Tb = BYTWJ(&(W[TWVL * 6]), Ta); cannam@167: Tc = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); cannam@167: Td = BYTWJ(&(W[0]), Tc); cannam@167: Te = VSUB(Tb, Td); cannam@167: Tk = VADD(Tb, Td); cannam@167: } cannam@167: { cannam@167: V T6, T8, T5, T7; cannam@167: T5 = LD(&(x[WS(rs, 2)]), ms, &(x[0])); cannam@167: T6 = BYTWJ(&(W[TWVL * 2]), T5); cannam@167: T7 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); cannam@167: T8 = BYTWJ(&(W[TWVL * 8]), T7); cannam@167: T9 = VSUB(T6, T8); cannam@167: Tj = VADD(T6, T8); cannam@167: } cannam@167: { cannam@167: V Th, Tf, Tg, Tn, Tl, Tm; cannam@167: Th = VMUL(LDK(KP866025403), VSUB(Te, T9)); cannam@167: Tf = VADD(T9, Te); cannam@167: Tg = VFNMS(LDK(KP500000000), Tf, T4); cannam@167: ST(&(x[WS(rs, 3)]), VADD(T4, Tf), ms, &(x[WS(rs, 1)])); cannam@167: ST(&(x[WS(rs, 1)]), VFMAI(Th, Tg), ms, &(x[WS(rs, 1)])); cannam@167: ST(&(x[WS(rs, 5)]), VFNMSI(Th, Tg), ms, &(x[WS(rs, 1)])); cannam@167: Tn = VMUL(LDK(KP866025403), VSUB(Tk, Tj)); cannam@167: Tl = VADD(Tj, Tk); cannam@167: Tm = VFNMS(LDK(KP500000000), Tl, Ti); cannam@167: ST(&(x[0]), VADD(Ti, Tl), ms, &(x[0])); cannam@167: ST(&(x[WS(rs, 4)]), VFMAI(Tn, Tm), ms, &(x[0])); cannam@167: ST(&(x[WS(rs, 2)]), VFNMSI(Tn, Tm), ms, &(x[0])); cannam@167: } cannam@167: } cannam@167: } cannam@167: VLEAVE(); cannam@167: } cannam@167: cannam@167: static const tw_instr twinstr[] = { cannam@167: VTW(0, 1), cannam@167: VTW(0, 2), cannam@167: VTW(0, 3), cannam@167: VTW(0, 4), cannam@167: VTW(0, 5), cannam@167: {TW_NEXT, VL, 0} cannam@167: }; cannam@167: cannam@167: static const ct_desc desc = { 6, XSIMD_STRING("t1fv_6"), twinstr, &GENUS, {17, 12, 6, 0}, 0, 0, 0 }; cannam@167: cannam@167: void XSIMD(codelet_t1fv_6) (planner *p) { cannam@167: X(kdft_dit_register) (p, t1fv_6, &desc); cannam@167: } cannam@167: #else cannam@167: cannam@167: /* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 6 -name t1fv_6 -include dft/simd/t1f.h */ cannam@167: cannam@167: /* cannam@167: * This function contains 23 FP additions, 14 FP multiplications, cannam@167: * (or, 21 additions, 12 multiplications, 2 fused multiply/add), cannam@167: * 19 stack variables, 2 constants, and 12 memory accesses cannam@167: */ cannam@167: #include "dft/simd/t1f.h" cannam@167: cannam@167: static void t1fv_6(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) cannam@167: { cannam@167: DVK(KP500000000, +0.500000000000000000000000000000000000000000000); cannam@167: DVK(KP866025403, +0.866025403784438646763723170752936183471402627); cannam@167: { cannam@167: INT m; cannam@167: R *x; cannam@167: x = ri; cannam@167: for (m = mb, W = W + (mb * ((TWVL / VL) * 10)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 10), MAKE_VOLATILE_STRIDE(6, rs)) { cannam@167: V T4, Ti, Te, Tk, T9, Tj, T1, T3, T2; cannam@167: T1 = LD(&(x[0]), ms, &(x[0])); cannam@167: T2 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); cannam@167: T3 = BYTWJ(&(W[TWVL * 4]), T2); cannam@167: T4 = VSUB(T1, T3); cannam@167: Ti = VADD(T1, T3); cannam@167: { cannam@167: V Tb, Td, Ta, Tc; cannam@167: Ta = LD(&(x[WS(rs, 4)]), ms, &(x[0])); cannam@167: Tb = BYTWJ(&(W[TWVL * 6]), Ta); cannam@167: Tc = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); cannam@167: Td = BYTWJ(&(W[0]), Tc); cannam@167: Te = VSUB(Tb, Td); cannam@167: Tk = VADD(Tb, Td); cannam@167: } cannam@167: { cannam@167: V T6, T8, T5, T7; cannam@167: T5 = LD(&(x[WS(rs, 2)]), ms, &(x[0])); cannam@167: T6 = BYTWJ(&(W[TWVL * 2]), T5); cannam@167: T7 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); cannam@167: T8 = BYTWJ(&(W[TWVL * 8]), T7); cannam@167: T9 = VSUB(T6, T8); cannam@167: Tj = VADD(T6, T8); cannam@167: } cannam@167: { cannam@167: V Th, Tf, Tg, Tn, Tl, Tm; cannam@167: Th = VBYI(VMUL(LDK(KP866025403), VSUB(Te, T9))); cannam@167: Tf = VADD(T9, Te); cannam@167: Tg = VFNMS(LDK(KP500000000), Tf, T4); cannam@167: ST(&(x[WS(rs, 3)]), VADD(T4, Tf), ms, &(x[WS(rs, 1)])); cannam@167: ST(&(x[WS(rs, 1)]), VADD(Tg, Th), ms, &(x[WS(rs, 1)])); cannam@167: ST(&(x[WS(rs, 5)]), VSUB(Tg, Th), ms, &(x[WS(rs, 1)])); cannam@167: Tn = VBYI(VMUL(LDK(KP866025403), VSUB(Tk, Tj))); cannam@167: Tl = VADD(Tj, Tk); cannam@167: Tm = VFNMS(LDK(KP500000000), Tl, Ti); cannam@167: ST(&(x[0]), VADD(Ti, Tl), ms, &(x[0])); cannam@167: ST(&(x[WS(rs, 4)]), VADD(Tm, Tn), ms, &(x[0])); cannam@167: ST(&(x[WS(rs, 2)]), VSUB(Tm, Tn), ms, &(x[0])); cannam@167: } cannam@167: } cannam@167: } cannam@167: VLEAVE(); cannam@167: } cannam@167: cannam@167: static const tw_instr twinstr[] = { cannam@167: VTW(0, 1), cannam@167: VTW(0, 2), cannam@167: VTW(0, 3), cannam@167: VTW(0, 4), cannam@167: VTW(0, 5), cannam@167: {TW_NEXT, VL, 0} cannam@167: }; cannam@167: cannam@167: static const ct_desc desc = { 6, XSIMD_STRING("t1fv_6"), twinstr, &GENUS, {21, 12, 2, 0}, 0, 0, 0 }; cannam@167: cannam@167: void XSIMD(codelet_t1fv_6) (planner *p) { cannam@167: X(kdft_dit_register) (p, t1fv_6, &desc); cannam@167: } cannam@167: #endif