Chris@10: /* Chris@10: * Copyright (c) 2003, 2007-11 Matteo Frigo Chris@10: * Copyright (c) 2003, 2007-11 Massachusetts Institute of Technology Chris@10: * Chris@10: * This program is free software; you can redistribute it and/or modify Chris@10: * it under the terms of the GNU General Public License as published by Chris@10: * the Free Software Foundation; either version 2 of the License, or Chris@10: * (at your option) any later version. Chris@10: * Chris@10: * This program is distributed in the hope that it will be useful, Chris@10: * but WITHOUT ANY WARRANTY; without even the implied warranty of Chris@10: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the Chris@10: * GNU General Public License for more details. Chris@10: * Chris@10: * You should have received a copy of the GNU General Public License Chris@10: * along with this program; if not, write to the Free Software Chris@10: * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA Chris@10: * Chris@10: */ Chris@10: Chris@10: /* This file was automatically generated --- DO NOT EDIT */ Chris@10: /* Generated on Sun Nov 25 07:40:30 EST 2012 */ Chris@10: Chris@10: #include "codelet-rdft.h" Chris@10: Chris@10: #ifdef HAVE_FMA Chris@10: Chris@10: /* Generated by: ../../../genfft/gen_hc2c.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 10 -dit -name hc2cf_10 -include hc2cf.h */ Chris@10: Chris@10: /* Chris@10: * This function contains 102 FP additions, 72 FP multiplications, Chris@10: * (or, 48 additions, 18 multiplications, 54 fused multiply/add), Chris@10: * 70 stack variables, 4 constants, and 40 memory accesses Chris@10: */ Chris@10: #include "hc2cf.h" Chris@10: Chris@10: static void hc2cf_10(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms) Chris@10: { Chris@10: DK(KP951056516, +0.951056516295153572116439333379382143405698634); Chris@10: DK(KP559016994, +0.559016994374947424102293417182819058860154590); Chris@10: DK(KP250000000, +0.250000000000000000000000000000000000000000000); Chris@10: DK(KP618033988, +0.618033988749894848204586834365638117720309180); Chris@10: { Chris@10: INT m; Chris@10: for (m = mb, W = W + ((mb - 1) * 18); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 18, MAKE_VOLATILE_STRIDE(40, rs)) { Chris@10: E T1X, T21, T20, T22; Chris@10: { Chris@10: E T26, T1U, T8, T12, T1n, T1P, T24, T1K, T1Y, T18, T10, T2b, T1H, T23, T15; Chris@10: E T1Z, T2a, Tz, T1O, T1y; Chris@10: { Chris@10: E T1, T1T, T3, T6, T2, T5; Chris@10: T1 = Rp[0]; Chris@10: T1T = Rm[0]; Chris@10: T3 = Ip[WS(rs, 2)]; Chris@10: T6 = Im[WS(rs, 2)]; Chris@10: T2 = W[8]; Chris@10: T5 = W[9]; Chris@10: { Chris@10: E T1l, TY, T1h, T1J, TM, T16, T1j, TS; Chris@10: { Chris@10: E TF, T1e, TO, TR, T1g, TL, TN, TQ, T1i, TP; Chris@10: { Chris@10: E TU, TX, TT, TW; Chris@10: { Chris@10: E TB, TE, T1R, T4, TA, TD; Chris@10: TB = Rp[WS(rs, 2)]; Chris@10: TE = Rm[WS(rs, 2)]; Chris@10: T1R = T2 * T6; Chris@10: T4 = T2 * T3; Chris@10: TA = W[6]; Chris@10: TD = W[7]; Chris@10: { Chris@10: E T1S, T7, T1d, TC; Chris@10: T1S = FNMS(T5, T3, T1R); Chris@10: T7 = FMA(T5, T6, T4); Chris@10: T1d = TA * TE; Chris@10: TC = TA * TB; Chris@10: T26 = T1T - T1S; Chris@10: T1U = T1S + T1T; Chris@10: T8 = T1 - T7; Chris@10: T12 = T1 + T7; Chris@10: TF = FMA(TD, TE, TC); Chris@10: T1e = FNMS(TD, TB, T1d); Chris@10: } Chris@10: } Chris@10: TU = Ip[0]; Chris@10: TX = Im[0]; Chris@10: TT = W[0]; Chris@10: TW = W[1]; Chris@10: { Chris@10: E TH, TK, TJ, T1f, TI, T1k, TV, TG; Chris@10: TH = Ip[WS(rs, 4)]; Chris@10: TK = Im[WS(rs, 4)]; Chris@10: T1k = TT * TX; Chris@10: TV = TT * TU; Chris@10: TG = W[16]; Chris@10: TJ = W[17]; Chris@10: T1l = FNMS(TW, TU, T1k); Chris@10: TY = FMA(TW, TX, TV); Chris@10: T1f = TG * TK; Chris@10: TI = TG * TH; Chris@10: TO = Rp[WS(rs, 3)]; Chris@10: TR = Rm[WS(rs, 3)]; Chris@10: T1g = FNMS(TJ, TH, T1f); Chris@10: TL = FMA(TJ, TK, TI); Chris@10: TN = W[10]; Chris@10: TQ = W[11]; Chris@10: } Chris@10: } Chris@10: T1h = T1e + T1g; Chris@10: T1J = T1g - T1e; Chris@10: TM = TF - TL; Chris@10: T16 = TF + TL; Chris@10: T1i = TN * TR; Chris@10: TP = TN * TO; Chris@10: T1j = FNMS(TQ, TO, T1i); Chris@10: TS = FMA(TQ, TR, TP); Chris@10: } Chris@10: { Chris@10: E T1p, Te, T1w, Tx, Tn, Tq, Tp, T1r, Tk, T1t, To; Chris@10: { Chris@10: E Tt, Tw, Tv, T1v, Tu; Chris@10: { Chris@10: E Ta, Td, T9, Tc, T1o, Tb, Ts; Chris@10: Ta = Rp[WS(rs, 1)]; Chris@10: Td = Rm[WS(rs, 1)]; Chris@10: { Chris@10: E T1I, T1m, TZ, T17; Chris@10: T1I = T1l - T1j; Chris@10: T1m = T1j + T1l; Chris@10: TZ = TS - TY; Chris@10: T17 = TS + TY; Chris@10: T1n = T1h - T1m; Chris@10: T1P = T1h + T1m; Chris@10: T24 = T1J + T1I; Chris@10: T1K = T1I - T1J; Chris@10: T1Y = T16 - T17; Chris@10: T18 = T16 + T17; Chris@10: T10 = TM + TZ; Chris@10: T2b = TZ - TM; Chris@10: T9 = W[2]; Chris@10: } Chris@10: Tc = W[3]; Chris@10: Tt = Ip[WS(rs, 1)]; Chris@10: Tw = Im[WS(rs, 1)]; Chris@10: T1o = T9 * Td; Chris@10: Tb = T9 * Ta; Chris@10: Ts = W[4]; Chris@10: Tv = W[5]; Chris@10: T1p = FNMS(Tc, Ta, T1o); Chris@10: Te = FMA(Tc, Td, Tb); Chris@10: T1v = Ts * Tw; Chris@10: Tu = Ts * Tt; Chris@10: } Chris@10: { Chris@10: E Tg, Tj, Tf, Ti, T1q, Th, Tm; Chris@10: Tg = Ip[WS(rs, 3)]; Chris@10: Tj = Im[WS(rs, 3)]; Chris@10: T1w = FNMS(Tv, Tt, T1v); Chris@10: Tx = FMA(Tv, Tw, Tu); Chris@10: Tf = W[12]; Chris@10: Ti = W[13]; Chris@10: Tn = Rp[WS(rs, 4)]; Chris@10: Tq = Rm[WS(rs, 4)]; Chris@10: T1q = Tf * Tj; Chris@10: Th = Tf * Tg; Chris@10: Tm = W[14]; Chris@10: Tp = W[15]; Chris@10: T1r = FNMS(Ti, Tg, T1q); Chris@10: Tk = FMA(Ti, Tj, Th); Chris@10: T1t = Tm * Tq; Chris@10: To = Tm * Tn; Chris@10: } Chris@10: } Chris@10: { Chris@10: E T1s, T1G, Tl, T13, T1u, Tr; Chris@10: T1s = T1p + T1r; Chris@10: T1G = T1r - T1p; Chris@10: Tl = Te - Tk; Chris@10: T13 = Te + Tk; Chris@10: T1u = FNMS(Tp, Tn, T1t); Chris@10: Tr = FMA(Tp, Tq, To); Chris@10: { Chris@10: E T1x, T1F, T14, Ty; Chris@10: T1x = T1u + T1w; Chris@10: T1F = T1w - T1u; Chris@10: T14 = Tr + Tx; Chris@10: Ty = Tr - Tx; Chris@10: T1H = T1F - T1G; Chris@10: T23 = T1G + T1F; Chris@10: T15 = T13 + T14; Chris@10: T1Z = T13 - T14; Chris@10: T2a = Ty - Tl; Chris@10: Tz = Tl + Ty; Chris@10: T1O = T1s + T1x; Chris@10: T1y = T1s - T1x; Chris@10: } Chris@10: } Chris@10: } Chris@10: } Chris@10: } Chris@10: { Chris@10: E T2c, T2e, T29, T2d; Chris@10: { Chris@10: E T1D, T11, T25, T28, T27; Chris@10: T1D = Tz - T10; Chris@10: T11 = Tz + T10; Chris@10: T25 = T23 + T24; Chris@10: T28 = T24 - T23; Chris@10: { Chris@10: E T1N, T1L, T1C, T1M, T1E; Chris@10: T1N = FNMS(KP618033988, T1H, T1K); Chris@10: T1L = FMA(KP618033988, T1K, T1H); Chris@10: Rm[WS(rs, 4)] = T8 + T11; Chris@10: T1C = FNMS(KP250000000, T11, T8); Chris@10: T1M = FNMS(KP559016994, T1D, T1C); Chris@10: T1E = FMA(KP559016994, T1D, T1C); Chris@10: T27 = FMA(KP250000000, T25, T26); Chris@10: T2c = FMA(KP618033988, T2b, T2a); Chris@10: T2e = FNMS(KP618033988, T2a, T2b); Chris@10: Rp[WS(rs, 1)] = FMA(KP951056516, T1L, T1E); Chris@10: Rm[0] = FNMS(KP951056516, T1L, T1E); Chris@10: Rp[WS(rs, 3)] = FMA(KP951056516, T1N, T1M); Chris@10: Rm[WS(rs, 2)] = FNMS(KP951056516, T1N, T1M); Chris@10: } Chris@10: Im[WS(rs, 4)] = T25 - T26; Chris@10: T29 = FMA(KP559016994, T28, T27); Chris@10: T2d = FNMS(KP559016994, T28, T27); Chris@10: } Chris@10: { Chris@10: E T1c, T1A, T1z, T1B, T19, T1b, T1a, T1Q, T1W, T1V; Chris@10: T19 = T15 + T18; Chris@10: T1b = T15 - T18; Chris@10: Ip[WS(rs, 3)] = FMA(KP951056516, T2e, T2d); Chris@10: Im[WS(rs, 2)] = FMS(KP951056516, T2e, T2d); Chris@10: Ip[WS(rs, 1)] = FMA(KP951056516, T2c, T29); Chris@10: Im[0] = FMS(KP951056516, T2c, T29); Chris@10: T1a = FNMS(KP250000000, T19, T12); Chris@10: Rp[0] = T12 + T19; Chris@10: T1c = FNMS(KP559016994, T1b, T1a); Chris@10: T1A = FMA(KP559016994, T1b, T1a); Chris@10: T1z = FNMS(KP618033988, T1y, T1n); Chris@10: T1B = FMA(KP618033988, T1n, T1y); Chris@10: T1Q = T1O + T1P; Chris@10: T1W = T1O - T1P; Chris@10: Rm[WS(rs, 3)] = FMA(KP951056516, T1B, T1A); Chris@10: Rp[WS(rs, 4)] = FNMS(KP951056516, T1B, T1A); Chris@10: Rm[WS(rs, 1)] = FMA(KP951056516, T1z, T1c); Chris@10: Rp[WS(rs, 2)] = FNMS(KP951056516, T1z, T1c); Chris@10: T1V = FNMS(KP250000000, T1Q, T1U); Chris@10: Ip[0] = T1Q + T1U; Chris@10: T1X = FNMS(KP559016994, T1W, T1V); Chris@10: T21 = FMA(KP559016994, T1W, T1V); Chris@10: T20 = FNMS(KP618033988, T1Z, T1Y); Chris@10: T22 = FMA(KP618033988, T1Y, T1Z); Chris@10: } Chris@10: } Chris@10: } Chris@10: Ip[WS(rs, 4)] = FMA(KP951056516, T22, T21); Chris@10: Im[WS(rs, 3)] = FMS(KP951056516, T22, T21); Chris@10: Ip[WS(rs, 2)] = FMA(KP951056516, T20, T1X); Chris@10: Im[WS(rs, 1)] = FMS(KP951056516, T20, T1X); Chris@10: } Chris@10: } Chris@10: } Chris@10: Chris@10: static const tw_instr twinstr[] = { Chris@10: {TW_FULL, 1, 10}, Chris@10: {TW_NEXT, 1, 0} Chris@10: }; Chris@10: Chris@10: static const hc2c_desc desc = { 10, "hc2cf_10", twinstr, &GENUS, {48, 18, 54, 0} }; Chris@10: Chris@10: void X(codelet_hc2cf_10) (planner *p) { Chris@10: X(khc2c_register) (p, hc2cf_10, &desc, HC2C_VIA_RDFT); Chris@10: } Chris@10: #else /* HAVE_FMA */ Chris@10: Chris@10: /* Generated by: ../../../genfft/gen_hc2c.native -compact -variables 4 -pipeline-latency 4 -n 10 -dit -name hc2cf_10 -include hc2cf.h */ Chris@10: Chris@10: /* Chris@10: * This function contains 102 FP additions, 60 FP multiplications, Chris@10: * (or, 72 additions, 30 multiplications, 30 fused multiply/add), Chris@10: * 45 stack variables, 4 constants, and 40 memory accesses Chris@10: */ Chris@10: #include "hc2cf.h" Chris@10: Chris@10: static void hc2cf_10(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms) Chris@10: { Chris@10: DK(KP587785252, +0.587785252292473129168705954639072768597652438); Chris@10: DK(KP951056516, +0.951056516295153572116439333379382143405698634); Chris@10: DK(KP250000000, +0.250000000000000000000000000000000000000000000); Chris@10: DK(KP559016994, +0.559016994374947424102293417182819058860154590); Chris@10: { Chris@10: INT m; Chris@10: for (m = mb, W = W + ((mb - 1) * 18); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 18, MAKE_VOLATILE_STRIDE(40, rs)) { Chris@10: E T7, T1O, TT, T1C, TF, TQ, TR, T1r, T1s, T1L, TX, TY, TZ, T16, T19; Chris@10: E T1y, Ti, Tt, Tu, T1o, T1p, T1M, TU, TV, TW, T1d, T1g, T1x; Chris@10: { Chris@10: E T1, T1B, T6, T1A; Chris@10: T1 = Rp[0]; Chris@10: T1B = Rm[0]; Chris@10: { Chris@10: E T3, T5, T2, T4; Chris@10: T3 = Ip[WS(rs, 2)]; Chris@10: T5 = Im[WS(rs, 2)]; Chris@10: T2 = W[8]; Chris@10: T4 = W[9]; Chris@10: T6 = FMA(T2, T3, T4 * T5); Chris@10: T1A = FNMS(T4, T3, T2 * T5); Chris@10: } Chris@10: T7 = T1 - T6; Chris@10: T1O = T1B - T1A; Chris@10: TT = T1 + T6; Chris@10: T1C = T1A + T1B; Chris@10: } Chris@10: { Chris@10: E Tz, T14, TP, T18, TE, T15, TK, T17; Chris@10: { Chris@10: E Tw, Ty, Tv, Tx; Chris@10: Tw = Rp[WS(rs, 2)]; Chris@10: Ty = Rm[WS(rs, 2)]; Chris@10: Tv = W[6]; Chris@10: Tx = W[7]; Chris@10: Tz = FMA(Tv, Tw, Tx * Ty); Chris@10: T14 = FNMS(Tx, Tw, Tv * Ty); Chris@10: } Chris@10: { Chris@10: E TM, TO, TL, TN; Chris@10: TM = Ip[0]; Chris@10: TO = Im[0]; Chris@10: TL = W[0]; Chris@10: TN = W[1]; Chris@10: TP = FMA(TL, TM, TN * TO); Chris@10: T18 = FNMS(TN, TM, TL * TO); Chris@10: } Chris@10: { Chris@10: E TB, TD, TA, TC; Chris@10: TB = Ip[WS(rs, 4)]; Chris@10: TD = Im[WS(rs, 4)]; Chris@10: TA = W[16]; Chris@10: TC = W[17]; Chris@10: TE = FMA(TA, TB, TC * TD); Chris@10: T15 = FNMS(TC, TB, TA * TD); Chris@10: } Chris@10: { Chris@10: E TH, TJ, TG, TI; Chris@10: TH = Rp[WS(rs, 3)]; Chris@10: TJ = Rm[WS(rs, 3)]; Chris@10: TG = W[10]; Chris@10: TI = W[11]; Chris@10: TK = FMA(TG, TH, TI * TJ); Chris@10: T17 = FNMS(TI, TH, TG * TJ); Chris@10: } Chris@10: TF = Tz - TE; Chris@10: TQ = TK - TP; Chris@10: TR = TF + TQ; Chris@10: T1r = T14 - T15; Chris@10: T1s = T18 - T17; Chris@10: T1L = T1s - T1r; Chris@10: TX = Tz + TE; Chris@10: TY = TK + TP; Chris@10: TZ = TX + TY; Chris@10: T16 = T14 + T15; Chris@10: T19 = T17 + T18; Chris@10: T1y = T16 + T19; Chris@10: } Chris@10: { Chris@10: E Tc, T1b, Ts, T1f, Th, T1c, Tn, T1e; Chris@10: { Chris@10: E T9, Tb, T8, Ta; Chris@10: T9 = Rp[WS(rs, 1)]; Chris@10: Tb = Rm[WS(rs, 1)]; Chris@10: T8 = W[2]; Chris@10: Ta = W[3]; Chris@10: Tc = FMA(T8, T9, Ta * Tb); Chris@10: T1b = FNMS(Ta, T9, T8 * Tb); Chris@10: } Chris@10: { Chris@10: E Tp, Tr, To, Tq; Chris@10: Tp = Ip[WS(rs, 1)]; Chris@10: Tr = Im[WS(rs, 1)]; Chris@10: To = W[4]; Chris@10: Tq = W[5]; Chris@10: Ts = FMA(To, Tp, Tq * Tr); Chris@10: T1f = FNMS(Tq, Tp, To * Tr); Chris@10: } Chris@10: { Chris@10: E Te, Tg, Td, Tf; Chris@10: Te = Ip[WS(rs, 3)]; Chris@10: Tg = Im[WS(rs, 3)]; Chris@10: Td = W[12]; Chris@10: Tf = W[13]; Chris@10: Th = FMA(Td, Te, Tf * Tg); Chris@10: T1c = FNMS(Tf, Te, Td * Tg); Chris@10: } Chris@10: { Chris@10: E Tk, Tm, Tj, Tl; Chris@10: Tk = Rp[WS(rs, 4)]; Chris@10: Tm = Rm[WS(rs, 4)]; Chris@10: Tj = W[14]; Chris@10: Tl = W[15]; Chris@10: Tn = FMA(Tj, Tk, Tl * Tm); Chris@10: T1e = FNMS(Tl, Tk, Tj * Tm); Chris@10: } Chris@10: Ti = Tc - Th; Chris@10: Tt = Tn - Ts; Chris@10: Tu = Ti + Tt; Chris@10: T1o = T1b - T1c; Chris@10: T1p = T1e - T1f; Chris@10: T1M = T1o + T1p; Chris@10: TU = Tc + Th; Chris@10: TV = Tn + Ts; Chris@10: TW = TU + TV; Chris@10: T1d = T1b + T1c; Chris@10: T1g = T1e + T1f; Chris@10: T1x = T1d + T1g; Chris@10: } Chris@10: { Chris@10: E T1l, TS, T1m, T1u, T1w, T1q, T1t, T1v, T1n; Chris@10: T1l = KP559016994 * (Tu - TR); Chris@10: TS = Tu + TR; Chris@10: T1m = FNMS(KP250000000, TS, T7); Chris@10: T1q = T1o - T1p; Chris@10: T1t = T1r + T1s; Chris@10: T1u = FMA(KP951056516, T1q, KP587785252 * T1t); Chris@10: T1w = FNMS(KP587785252, T1q, KP951056516 * T1t); Chris@10: Rm[WS(rs, 4)] = T7 + TS; Chris@10: T1v = T1m - T1l; Chris@10: Rm[WS(rs, 2)] = T1v - T1w; Chris@10: Rp[WS(rs, 3)] = T1v + T1w; Chris@10: T1n = T1l + T1m; Chris@10: Rm[0] = T1n - T1u; Chris@10: Rp[WS(rs, 1)] = T1n + T1u; Chris@10: } Chris@10: { Chris@10: E T1S, T1N, T1T, T1R, T1V, T1P, T1Q, T1W, T1U; Chris@10: T1S = KP559016994 * (T1M + T1L); Chris@10: T1N = T1L - T1M; Chris@10: T1T = FMA(KP250000000, T1N, T1O); Chris@10: T1P = TQ - TF; Chris@10: T1Q = Ti - Tt; Chris@10: T1R = FNMS(KP951056516, T1Q, KP587785252 * T1P); Chris@10: T1V = FMA(KP587785252, T1Q, KP951056516 * T1P); Chris@10: Im[WS(rs, 4)] = T1N - T1O; Chris@10: T1W = T1T - T1S; Chris@10: Im[WS(rs, 2)] = T1V - T1W; Chris@10: Ip[WS(rs, 3)] = T1V + T1W; Chris@10: T1U = T1S + T1T; Chris@10: Im[0] = T1R - T1U; Chris@10: Ip[WS(rs, 1)] = T1R + T1U; Chris@10: } Chris@10: { Chris@10: E T12, T10, T11, T1i, T1k, T1a, T1h, T1j, T13; Chris@10: T12 = KP559016994 * (TW - TZ); Chris@10: T10 = TW + TZ; Chris@10: T11 = FNMS(KP250000000, T10, TT); Chris@10: T1a = T16 - T19; Chris@10: T1h = T1d - T1g; Chris@10: T1i = FNMS(KP587785252, T1h, KP951056516 * T1a); Chris@10: T1k = FMA(KP951056516, T1h, KP587785252 * T1a); Chris@10: Rp[0] = TT + T10; Chris@10: T1j = T12 + T11; Chris@10: Rp[WS(rs, 4)] = T1j - T1k; Chris@10: Rm[WS(rs, 3)] = T1j + T1k; Chris@10: T13 = T11 - T12; Chris@10: Rp[WS(rs, 2)] = T13 - T1i; Chris@10: Rm[WS(rs, 1)] = T13 + T1i; Chris@10: } Chris@10: { Chris@10: E T1H, T1z, T1G, T1F, T1J, T1D, T1E, T1K, T1I; Chris@10: T1H = KP559016994 * (T1x - T1y); Chris@10: T1z = T1x + T1y; Chris@10: T1G = FNMS(KP250000000, T1z, T1C); Chris@10: T1D = TX - TY; Chris@10: T1E = TU - TV; Chris@10: T1F = FNMS(KP587785252, T1E, KP951056516 * T1D); Chris@10: T1J = FMA(KP951056516, T1E, KP587785252 * T1D); Chris@10: Ip[0] = T1z + T1C; Chris@10: T1K = T1H + T1G; Chris@10: Im[WS(rs, 3)] = T1J - T1K; Chris@10: Ip[WS(rs, 4)] = T1J + T1K; Chris@10: T1I = T1G - T1H; Chris@10: Im[WS(rs, 1)] = T1F - T1I; Chris@10: Ip[WS(rs, 2)] = T1F + T1I; Chris@10: } Chris@10: } Chris@10: } Chris@10: } Chris@10: Chris@10: static const tw_instr twinstr[] = { Chris@10: {TW_FULL, 1, 10}, Chris@10: {TW_NEXT, 1, 0} Chris@10: }; Chris@10: Chris@10: static const hc2c_desc desc = { 10, "hc2cf_10", twinstr, &GENUS, {72, 30, 30, 0} }; Chris@10: Chris@10: void X(codelet_hc2cf_10) (planner *p) { Chris@10: X(khc2c_register) (p, hc2cf_10, &desc, HC2C_VIA_RDFT); Chris@10: } Chris@10: #endif /* HAVE_FMA */