summaryrefslogtreecommitdiff
path: root/src/fftw3/rdft/codelets/hc2r/hb_12.c
diff options
context:
space:
mode:
Diffstat (limited to 'src/fftw3/rdft/codelets/hc2r/hb_12.c')
-rw-r--r--src/fftw3/rdft/codelets/hc2r/hb_12.c299
1 files changed, 299 insertions, 0 deletions
diff --git a/src/fftw3/rdft/codelets/hc2r/hb_12.c b/src/fftw3/rdft/codelets/hc2r/hb_12.c
new file mode 100644
index 0000000..55e7db0
--- /dev/null
+++ b/src/fftw3/rdft/codelets/hc2r/hb_12.c
@@ -0,0 +1,299 @@
+/*
+ * Copyright (c) 2003 Matteo Frigo
+ * Copyright (c) 2003 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 Sat Jul 5 22:11:36 EDT 2003 */
+
+#include "codelet-rdft.h"
+
+/* Generated by: /homee/stevenj/cvs/fftw3.0.1/genfft/gen_hc2hc -compact -variables 4 -sign 1 -n 12 -dif -name hb_12 -include hb.h */
+
+/*
+ * This function contains 118 FP additions, 60 FP multiplications,
+ * (or, 88 additions, 30 multiplications, 30 fused multiply/add),
+ * 39 stack variables, and 48 memory accesses
+ */
+/*
+ * Generator Id's :
+ * $Id: hb_12.c,v 1.1 2008/10/17 06:12:08 scuri Exp $
+ * $Id: hb_12.c,v 1.1 2008/10/17 06:12:08 scuri Exp $
+ * $Id: hb_12.c,v 1.1 2008/10/17 06:12:08 scuri Exp $
+ */
+
+#include "hb.h"
+
+static const R *hb_12(R *rio, R *iio, const R *W, stride ios, int m, int dist)
+{
+ DK(KP500000000, +0.500000000000000000000000000000000000000000000);
+ DK(KP866025403, +0.866025403784438646763723170752936183471402627);
+ int i;
+ for (i = m - 2; i > 0; i = i - 2, rio = rio + dist, iio = iio - dist, W = W + 22) {
+ E T5, Tt, T12, T1M, T1i, T1U, Tl, TM, T1c, T1Y, T1s, T1Q, Ta, Ty, T15;
+ E T1N, T1l, T1V, Tg, TH, T19, T1X, T1p, T1P;
+ {
+ E T1, Tp, T4, T1g, Ts, T11, T10, T1h;
+ T1 = rio[0];
+ Tp = iio[0];
+ {
+ E T2, T3, Tq, Tr;
+ T2 = rio[WS(ios, 4)];
+ T3 = iio[-WS(ios, 8)];
+ T4 = T2 + T3;
+ T1g = KP866025403 * (T2 - T3);
+ Tq = rio[WS(ios, 8)];
+ Tr = iio[-WS(ios, 4)];
+ Ts = Tq - Tr;
+ T11 = KP866025403 * (Tq + Tr);
+ }
+ T5 = T1 + T4;
+ Tt = Tp - Ts;
+ T10 = FNMS(KP500000000, T4, T1);
+ T12 = T10 - T11;
+ T1M = T10 + T11;
+ T1h = FMA(KP500000000, Ts, Tp);
+ T1i = T1g + T1h;
+ T1U = T1h - T1g;
+ }
+ {
+ E Th, TL, Tk, T1a, TK, T1r, T1b, T1q;
+ Th = iio[-WS(ios, 9)];
+ TL = rio[WS(ios, 9)];
+ {
+ E Ti, Tj, TI, TJ;
+ Ti = rio[WS(ios, 1)];
+ Tj = rio[WS(ios, 5)];
+ Tk = Ti + Tj;
+ T1a = KP866025403 * (Ti - Tj);
+ TI = iio[-WS(ios, 5)];
+ TJ = iio[-WS(ios, 1)];
+ TK = TI + TJ;
+ T1r = KP866025403 * (TI - TJ);
+ }
+ Tl = Th + Tk;
+ TM = TK - TL;
+ T1b = FMA(KP500000000, TK, TL);
+ T1c = T1a - T1b;
+ T1Y = T1a + T1b;
+ T1q = FNMS(KP500000000, Tk, Th);
+ T1s = T1q + T1r;
+ T1Q = T1q - T1r;
+ }
+ {
+ E T6, Tx, T9, T1j, Tw, T14, T13, T1k;
+ T6 = iio[-WS(ios, 6)];
+ Tx = rio[WS(ios, 6)];
+ {
+ E T7, T8, Tu, Tv;
+ T7 = iio[-WS(ios, 10)];
+ T8 = rio[WS(ios, 2)];
+ T9 = T7 + T8;
+ T1j = KP866025403 * (T7 - T8);
+ Tu = rio[WS(ios, 10)];
+ Tv = iio[-WS(ios, 2)];
+ Tw = Tu - Tv;
+ T14 = KP866025403 * (Tu + Tv);
+ }
+ Ta = T6 + T9;
+ Ty = Tw + Tx;
+ T13 = FNMS(KP500000000, T9, T6);
+ T15 = T13 + T14;
+ T1N = T13 - T14;
+ T1k = FMS(KP500000000, Tw, Tx);
+ T1l = T1j + T1k;
+ T1V = T1k - T1j;
+ }
+ {
+ E Tc, TD, Tf, T17, TG, T1o, T18, T1n;
+ Tc = rio[WS(ios, 3)];
+ TD = iio[-WS(ios, 3)];
+ {
+ E Td, Te, TE, TF;
+ Td = iio[-WS(ios, 7)];
+ Te = iio[-WS(ios, 11)];
+ Tf = Td + Te;
+ T17 = KP866025403 * (Td - Te);
+ TE = rio[WS(ios, 7)];
+ TF = rio[WS(ios, 11)];
+ TG = TE + TF;
+ T1o = KP866025403 * (TE - TF);
+ }
+ Tg = Tc + Tf;
+ TH = TD - TG;
+ T18 = FMA(KP500000000, TG, TD);
+ T19 = T17 + T18;
+ T1X = T18 - T17;
+ T1n = FNMS(KP500000000, Tf, Tc);
+ T1p = T1n + T1o;
+ T1P = T1n - T1o;
+ }
+ {
+ E Tb, Tm, TU, TW, TX, TY, TT, TV;
+ Tb = T5 + Ta;
+ Tm = Tg + Tl;
+ TU = Tb - Tm;
+ TW = Tt - Ty;
+ TX = TH + TM;
+ TY = TW - TX;
+ rio[0] = Tb + Tm;
+ iio[-WS(ios, 11)] = TW + TX;
+ TT = W[10];
+ TV = W[11];
+ rio[WS(ios, 6)] = FNMS(TV, TY, TT * TU);
+ iio[-WS(ios, 5)] = FMA(TV, TU, TT * TY);
+ }
+ {
+ E T28, T2g, T2c, T2e;
+ {
+ E T26, T27, T2a, T2b;
+ T26 = T1M - T1N;
+ T27 = T1X + T1Y;
+ T28 = T26 - T27;
+ T2g = T26 + T27;
+ T2a = T1U - T1V;
+ T2b = T1P - T1Q;
+ T2c = T2a + T2b;
+ T2e = T2a - T2b;
+ }
+ {
+ E T25, T29, T2d, T2f;
+ T25 = W[8];
+ T29 = W[9];
+ rio[WS(ios, 5)] = FNMS(T29, T2c, T25 * T28);
+ iio[-WS(ios, 6)] = FMA(T25, T2c, T29 * T28);
+ T2d = W[20];
+ T2f = W[21];
+ iio[0] = FMA(T2d, T2e, T2f * T2g);
+ rio[WS(ios, 11)] = FNMS(T2f, T2e, T2d * T2g);
+ }
+ }
+ {
+ E TA, TS, TO, TQ;
+ {
+ E To, Tz, TC, TN;
+ To = Tg - Tl;
+ Tz = Tt + Ty;
+ TA = To + Tz;
+ TS = Tz - To;
+ TC = T5 - Ta;
+ TN = TH - TM;
+ TO = TC - TN;
+ TQ = TC + TN;
+ }
+ {
+ E Tn, TB, TP, TR;
+ Tn = W[16];
+ TB = W[17];
+ iio[-WS(ios, 2)] = FMA(Tn, TA, TB * TO);
+ rio[WS(ios, 9)] = FNMS(TB, TA, Tn * TO);
+ TP = W[4];
+ TR = W[5];
+ rio[WS(ios, 3)] = FNMS(TR, TS, TP * TQ);
+ iio[-WS(ios, 8)] = FMA(TP, TS, TR * TQ);
+ }
+ }
+ {
+ E T1S, T22, T20, T24;
+ {
+ E T1O, T1R, T1W, T1Z;
+ T1O = T1M + T1N;
+ T1R = T1P + T1Q;
+ T1S = T1O - T1R;
+ T22 = T1O + T1R;
+ T1W = T1U + T1V;
+ T1Z = T1X - T1Y;
+ T20 = T1W - T1Z;
+ T24 = T1W + T1Z;
+ }
+ {
+ E T1L, T1T, T21, T23;
+ T1L = W[2];
+ T1T = W[3];
+ rio[WS(ios, 2)] = FNMS(T1T, T20, T1L * T1S);
+ iio[-WS(ios, 9)] = FMA(T1T, T1S, T1L * T20);
+ T21 = W[14];
+ T23 = W[15];
+ rio[WS(ios, 8)] = FNMS(T23, T24, T21 * T22);
+ iio[-WS(ios, 3)] = FMA(T23, T22, T21 * T24);
+ }
+ }
+ {
+ E T1C, T1I, T1G, T1K;
+ {
+ E T1A, T1B, T1E, T1F;
+ T1A = T12 + T15;
+ T1B = T1p + T1s;
+ T1C = T1A - T1B;
+ T1I = T1A + T1B;
+ T1E = T1i + T1l;
+ T1F = T19 + T1c;
+ T1G = T1E - T1F;
+ T1K = T1E + T1F;
+ }
+ {
+ E T1z, T1D, T1H, T1J;
+ T1z = W[18];
+ T1D = W[19];
+ rio[WS(ios, 10)] = FNMS(T1D, T1G, T1z * T1C);
+ iio[-WS(ios, 1)] = FMA(T1D, T1C, T1z * T1G);
+ T1H = W[6];
+ T1J = W[7];
+ rio[WS(ios, 4)] = FNMS(T1J, T1K, T1H * T1I);
+ iio[-WS(ios, 7)] = FMA(T1J, T1I, T1H * T1K);
+ }
+ }
+ {
+ E T1e, T1y, T1u, T1w;
+ {
+ E T16, T1d, T1m, T1t;
+ T16 = T12 - T15;
+ T1d = T19 - T1c;
+ T1e = T16 - T1d;
+ T1y = T16 + T1d;
+ T1m = T1i - T1l;
+ T1t = T1p - T1s;
+ T1u = T1m + T1t;
+ T1w = T1m - T1t;
+ }
+ {
+ E TZ, T1f, T1v, T1x;
+ TZ = W[0];
+ T1f = W[1];
+ rio[WS(ios, 1)] = FNMS(T1f, T1u, TZ * T1e);
+ iio[-WS(ios, 10)] = FMA(TZ, T1u, T1f * T1e);
+ T1v = W[12];
+ T1x = W[13];
+ iio[-WS(ios, 4)] = FMA(T1v, T1w, T1x * T1y);
+ rio[WS(ios, 7)] = FNMS(T1x, T1w, T1v * T1y);
+ }
+ }
+ }
+ return W;
+}
+
+static const tw_instr twinstr[] = {
+ {TW_FULL, 0, 12},
+ {TW_NEXT, 1, 0}
+};
+
+static const hc2hc_desc desc = { 12, "hb_12", twinstr, {88, 30, 30, 0}, &GENUS, 0, 0, 0 };
+
+void X(codelet_hb_12) (planner *p) {
+ X(khc2hc_dif_register) (p, hb_12, &desc);
+}