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 13:46:52 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_notw_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 14 -name n1bv_14 -include n1b.h */ Chris@19: Chris@19: /* Chris@19: * This function contains 74 FP additions, 48 FP multiplications, Chris@19: * (or, 32 additions, 6 multiplications, 42 fused multiply/add), Chris@19: * 63 stack variables, 6 constants, and 28 memory accesses Chris@19: */ Chris@19: #include "n1b.h" Chris@19: Chris@19: static void n1bv_14(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) Chris@19: { Chris@19: DVK(KP900968867, +0.900968867902419126236102319507445051165919162); Chris@19: DVK(KP801937735, +0.801937735804838252472204639014890102331838324); Chris@19: DVK(KP974927912, +0.974927912181823607018131682993931217232785801); Chris@19: DVK(KP692021471, +0.692021471630095869627814897002069140197260599); Chris@19: DVK(KP554958132, +0.554958132087371191422194871006410481067288862); Chris@19: DVK(KP356895867, +0.356895867892209443894399510021300583399127187); Chris@19: { Chris@19: INT i; Chris@19: const R *xi; Chris@19: R *xo; Chris@19: xi = ii; Chris@19: xo = io; Chris@19: for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(28, is), MAKE_VOLATILE_STRIDE(28, os)) { Chris@19: V TH, T3, TP, Tn, Ta, Tu, TU, TK, TO, Tk, TM, Tg, TL, Td, T1; Chris@19: V T2; Chris@19: T1 = LD(&(xi[0]), ivs, &(xi[0])); Chris@19: T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Chris@19: { Chris@19: V Ti, TI, T6, TJ, T9, Tj, Te, Tf, Tb, Tc; Chris@19: { Chris@19: V T4, T5, T7, T8, Tl, Tm; Chris@19: T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Chris@19: T5 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); Chris@19: T7 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); Chris@19: T8 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Chris@19: Tl = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); Chris@19: Tm = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); Chris@19: Ti = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); Chris@19: TH = VADD(T1, T2); Chris@19: T3 = VSUB(T1, T2); Chris@19: TI = VADD(T4, T5); Chris@19: T6 = VSUB(T4, T5); Chris@19: TJ = VADD(T7, T8); Chris@19: T9 = VSUB(T7, T8); Chris@19: TP = VADD(Tl, Tm); Chris@19: Tn = VSUB(Tl, Tm); Chris@19: Tj = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); Chris@19: Te = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); Chris@19: Tf = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Chris@19: Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); Chris@19: Tc = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); Chris@19: } Chris@19: Ta = VADD(T6, T9); Chris@19: Tu = VSUB(T6, T9); Chris@19: TU = VSUB(TI, TJ); Chris@19: TK = VADD(TI, TJ); Chris@19: TO = VADD(Ti, Tj); Chris@19: Tk = VSUB(Ti, Tj); Chris@19: TM = VADD(Te, Tf); Chris@19: Tg = VSUB(Te, Tf); Chris@19: TL = VADD(Tb, Tc); Chris@19: Td = VSUB(Tb, Tc); Chris@19: } Chris@19: { Chris@19: V T13, TG, TY, T18, TB, Tw, TT, Tz, T11, T16, TE, Tr, TV, TQ; Chris@19: TV = VSUB(TP, TO); Chris@19: TQ = VADD(TO, TP); Chris@19: { Chris@19: V Ts, To, TW, TN; Chris@19: Ts = VSUB(Tk, Tn); Chris@19: To = VADD(Tk, Tn); Chris@19: TW = VSUB(TM, TL); Chris@19: TN = VADD(TL, TM); Chris@19: { Chris@19: V Tt, Th, TR, T12; Chris@19: Tt = VSUB(Td, Tg); Chris@19: Th = VADD(Td, Tg); Chris@19: TR = VFNMS(LDK(KP356895867), TK, TQ); Chris@19: T12 = VFNMS(LDK(KP554958132), TV, TU); Chris@19: { Chris@19: V Tx, TF, TZ, T14; Chris@19: Tx = VFNMS(LDK(KP356895867), Ta, To); Chris@19: TF = VFMA(LDK(KP554958132), Ts, Tu); Chris@19: ST(&(xo[0]), VADD(TH, VADD(TK, VADD(TN, TQ))), ovs, &(xo[0])); Chris@19: TZ = VFNMS(LDK(KP356895867), TN, TK); Chris@19: T14 = VFNMS(LDK(KP356895867), TQ, TN); Chris@19: { Chris@19: V TX, T17, TC, Tp; Chris@19: TX = VFMA(LDK(KP554958132), TW, TV); Chris@19: T17 = VFMA(LDK(KP554958132), TU, TW); Chris@19: ST(&(xo[WS(os, 7)]), VADD(T3, VADD(Ta, VADD(Th, To))), ovs, &(xo[WS(os, 1)])); Chris@19: TC = VFNMS(LDK(KP356895867), Th, Ta); Chris@19: Tp = VFNMS(LDK(KP356895867), To, Th); Chris@19: { Chris@19: V TA, Tv, TS, Ty; Chris@19: TA = VFMA(LDK(KP554958132), Tt, Ts); Chris@19: Tv = VFNMS(LDK(KP554958132), Tu, Tt); Chris@19: TS = VFNMS(LDK(KP692021471), TR, TN); Chris@19: T13 = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), T12, TW)); Chris@19: Ty = VFNMS(LDK(KP692021471), Tx, Th); Chris@19: TG = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), TF, Tt)); Chris@19: { Chris@19: V T10, T15, TD, Tq; Chris@19: T10 = VFNMS(LDK(KP692021471), TZ, TQ); Chris@19: T15 = VFNMS(LDK(KP692021471), T14, TK); Chris@19: TY = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), TX, TU)); Chris@19: T18 = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), T17, TV)); Chris@19: TD = VFNMS(LDK(KP692021471), TC, To); Chris@19: Tq = VFNMS(LDK(KP692021471), Tp, Ta); Chris@19: TB = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), TA, Tu)); Chris@19: Tw = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tv, Ts)); Chris@19: TT = VFNMS(LDK(KP900968867), TS, TH); Chris@19: Tz = VFNMS(LDK(KP900968867), Ty, T3); Chris@19: T11 = VFNMS(LDK(KP900968867), T10, TH); Chris@19: T16 = VFNMS(LDK(KP900968867), T15, TH); Chris@19: TE = VFNMS(LDK(KP900968867), TD, T3); Chris@19: Tr = VFNMS(LDK(KP900968867), Tq, T3); Chris@19: } Chris@19: } Chris@19: } Chris@19: } Chris@19: } Chris@19: } Chris@19: ST(&(xo[WS(os, 2)]), VFMAI(TY, TT), ovs, &(xo[0])); Chris@19: ST(&(xo[WS(os, 12)]), VFNMSI(TY, TT), ovs, &(xo[0])); Chris@19: ST(&(xo[WS(os, 9)]), VFMAI(TB, Tz), ovs, &(xo[WS(os, 1)])); Chris@19: ST(&(xo[WS(os, 5)]), VFNMSI(TB, Tz), ovs, &(xo[WS(os, 1)])); Chris@19: ST(&(xo[WS(os, 6)]), VFMAI(T13, T11), ovs, &(xo[0])); Chris@19: ST(&(xo[WS(os, 8)]), VFNMSI(T13, T11), ovs, &(xo[0])); Chris@19: ST(&(xo[WS(os, 4)]), VFMAI(T18, T16), ovs, &(xo[0])); Chris@19: ST(&(xo[WS(os, 10)]), VFNMSI(T18, T16), ovs, &(xo[0])); Chris@19: ST(&(xo[WS(os, 13)]), VFNMSI(TG, TE), ovs, &(xo[WS(os, 1)])); Chris@19: ST(&(xo[WS(os, 1)]), VFMAI(TG, TE), ovs, &(xo[WS(os, 1)])); Chris@19: ST(&(xo[WS(os, 11)]), VFNMSI(Tw, Tr), ovs, &(xo[WS(os, 1)])); Chris@19: ST(&(xo[WS(os, 3)]), VFMAI(Tw, Tr), ovs, &(xo[WS(os, 1)])); Chris@19: } Chris@19: } Chris@19: } Chris@19: VLEAVE(); Chris@19: } Chris@19: Chris@19: static const kdft_desc desc = { 14, XSIMD_STRING("n1bv_14"), {32, 6, 42, 0}, &GENUS, 0, 0, 0, 0 }; Chris@19: Chris@19: void XSIMD(codelet_n1bv_14) (planner *p) { Chris@19: X(kdft_register) (p, n1bv_14, &desc); Chris@19: } Chris@19: Chris@19: #else /* HAVE_FMA */ Chris@19: Chris@19: /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 14 -name n1bv_14 -include n1b.h */ Chris@19: Chris@19: /* Chris@19: * This function contains 74 FP additions, 36 FP multiplications, Chris@19: * (or, 50 additions, 12 multiplications, 24 fused multiply/add), Chris@19: * 33 stack variables, 6 constants, and 28 memory accesses Chris@19: */ Chris@19: #include "n1b.h" Chris@19: Chris@19: static void n1bv_14(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) Chris@19: { Chris@19: DVK(KP900968867, +0.900968867902419126236102319507445051165919162); Chris@19: DVK(KP222520933, +0.222520933956314404288902564496794759466355569); Chris@19: DVK(KP623489801, +0.623489801858733530525004884004239810632274731); Chris@19: DVK(KP781831482, +0.781831482468029808708444526674057750232334519); Chris@19: DVK(KP974927912, +0.974927912181823607018131682993931217232785801); Chris@19: DVK(KP433883739, +0.433883739117558120475768332848358754609990728); Chris@19: { Chris@19: INT i; Chris@19: const R *xi; Chris@19: R *xo; Chris@19: xi = ii; Chris@19: xo = io; Chris@19: for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(28, is), MAKE_VOLATILE_STRIDE(28, os)) { Chris@19: V Tp, Ty, Tl, TL, Tq, TE, T7, TJ, Ts, TB, Te, TK, Tr, TH, Tn; Chris@19: V To; Chris@19: Tn = LD(&(xi[0]), ivs, &(xi[0])); Chris@19: To = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Chris@19: Tp = VSUB(Tn, To); Chris@19: Ty = VADD(Tn, To); Chris@19: { Chris@19: V Th, TC, Tk, TD; Chris@19: { Chris@19: V Tf, Tg, Ti, Tj; Chris@19: Tf = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); Chris@19: Tg = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); Chris@19: Th = VSUB(Tf, Tg); Chris@19: TC = VADD(Tf, Tg); Chris@19: Ti = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); Chris@19: Tj = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Chris@19: Tk = VSUB(Ti, Tj); Chris@19: TD = VADD(Ti, Tj); Chris@19: } Chris@19: Tl = VSUB(Th, Tk); Chris@19: TL = VSUB(TD, TC); Chris@19: Tq = VADD(Th, Tk); Chris@19: TE = VADD(TC, TD); Chris@19: } Chris@19: { Chris@19: V T3, Tz, T6, TA; Chris@19: { Chris@19: V T1, T2, T4, T5; Chris@19: T1 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Chris@19: T2 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); Chris@19: T3 = VSUB(T1, T2); Chris@19: Tz = VADD(T1, T2); Chris@19: T4 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); Chris@19: T5 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Chris@19: T6 = VSUB(T4, T5); Chris@19: TA = VADD(T4, T5); Chris@19: } Chris@19: T7 = VSUB(T3, T6); Chris@19: TJ = VSUB(Tz, TA); Chris@19: Ts = VADD(T3, T6); Chris@19: TB = VADD(Tz, TA); Chris@19: } Chris@19: { Chris@19: V Ta, TF, Td, TG; Chris@19: { Chris@19: V T8, T9, Tb, Tc; Chris@19: T8 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); Chris@19: T9 = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); Chris@19: Ta = VSUB(T8, T9); Chris@19: TF = VADD(T8, T9); Chris@19: Tb = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); Chris@19: Tc = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); Chris@19: Td = VSUB(Tb, Tc); Chris@19: TG = VADD(Tb, Tc); Chris@19: } Chris@19: Te = VSUB(Ta, Td); Chris@19: TK = VSUB(TG, TF); Chris@19: Tr = VADD(Ta, Td); Chris@19: TH = VADD(TF, TG); Chris@19: } Chris@19: ST(&(xo[WS(os, 7)]), VADD(Tp, VADD(Ts, VADD(Tq, Tr))), ovs, &(xo[WS(os, 1)])); Chris@19: ST(&(xo[0]), VADD(Ty, VADD(TB, VADD(TE, TH))), ovs, &(xo[0])); Chris@19: { Chris@19: V Tm, Tt, TQ, TP; Chris@19: Tm = VBYI(VFMA(LDK(KP433883739), T7, VFNMS(LDK(KP781831482), Tl, VMUL(LDK(KP974927912), Te)))); Chris@19: Tt = VFMA(LDK(KP623489801), Tq, VFNMS(LDK(KP222520933), Tr, VFNMS(LDK(KP900968867), Ts, Tp))); Chris@19: ST(&(xo[WS(os, 3)]), VADD(Tm, Tt), ovs, &(xo[WS(os, 1)])); Chris@19: ST(&(xo[WS(os, 11)]), VSUB(Tt, Tm), ovs, &(xo[WS(os, 1)])); Chris@19: TQ = VBYI(VFMA(LDK(KP974927912), TJ, VFMA(LDK(KP433883739), TL, VMUL(LDK(KP781831482), TK)))); Chris@19: TP = VFMA(LDK(KP623489801), TH, VFNMS(LDK(KP900968867), TE, VFNMS(LDK(KP222520933), TB, Ty))); Chris@19: ST(&(xo[WS(os, 12)]), VSUB(TP, TQ), ovs, &(xo[0])); Chris@19: ST(&(xo[WS(os, 2)]), VADD(TP, TQ), ovs, &(xo[0])); Chris@19: } Chris@19: { Chris@19: V Tu, Tv, TM, TI; Chris@19: Tu = VBYI(VFMA(LDK(KP781831482), T7, VFMA(LDK(KP974927912), Tl, VMUL(LDK(KP433883739), Te)))); Chris@19: Tv = VFMA(LDK(KP623489801), Ts, VFNMS(LDK(KP900968867), Tr, VFNMS(LDK(KP222520933), Tq, Tp))); Chris@19: ST(&(xo[WS(os, 1)]), VADD(Tu, Tv), ovs, &(xo[WS(os, 1)])); Chris@19: ST(&(xo[WS(os, 13)]), VSUB(Tv, Tu), ovs, &(xo[WS(os, 1)])); Chris@19: TM = VBYI(VFNMS(LDK(KP433883739), TK, VFNMS(LDK(KP974927912), TL, VMUL(LDK(KP781831482), TJ)))); Chris@19: TI = VFMA(LDK(KP623489801), TB, VFNMS(LDK(KP900968867), TH, VFNMS(LDK(KP222520933), TE, Ty))); Chris@19: ST(&(xo[WS(os, 6)]), VSUB(TI, TM), ovs, &(xo[0])); Chris@19: ST(&(xo[WS(os, 8)]), VADD(TI, TM), ovs, &(xo[0])); Chris@19: } Chris@19: { Chris@19: V TO, TN, Tx, Tw; Chris@19: TO = VBYI(VFMA(LDK(KP433883739), TJ, VFNMS(LDK(KP974927912), TK, VMUL(LDK(KP781831482), TL)))); Chris@19: TN = VFMA(LDK(KP623489801), TE, VFNMS(LDK(KP222520933), TH, VFNMS(LDK(KP900968867), TB, Ty))); Chris@19: ST(&(xo[WS(os, 4)]), VSUB(TN, TO), ovs, &(xo[0])); Chris@19: ST(&(xo[WS(os, 10)]), VADD(TN, TO), ovs, &(xo[0])); Chris@19: Tx = VBYI(VFNMS(LDK(KP781831482), Te, VFNMS(LDK(KP433883739), Tl, VMUL(LDK(KP974927912), T7)))); Chris@19: Tw = VFMA(LDK(KP623489801), Tr, VFNMS(LDK(KP900968867), Tq, VFNMS(LDK(KP222520933), Ts, Tp))); Chris@19: ST(&(xo[WS(os, 5)]), VSUB(Tw, Tx), ovs, &(xo[WS(os, 1)])); Chris@19: ST(&(xo[WS(os, 9)]), VADD(Tx, Tw), ovs, &(xo[WS(os, 1)])); Chris@19: } Chris@19: } Chris@19: } Chris@19: VLEAVE(); Chris@19: } Chris@19: Chris@19: static const kdft_desc desc = { 14, XSIMD_STRING("n1bv_14"), {50, 12, 24, 0}, &GENUS, 0, 0, 0, 0 }; Chris@19: Chris@19: void XSIMD(codelet_n1bv_14) (planner *p) { Chris@19: X(kdft_register) (p, n1bv_14, &desc); Chris@19: } Chris@19: Chris@19: #endif /* HAVE_FMA */