summaryrefslogtreecommitdiff
path: root/src/fftw/fni_4.c
diff options
context:
space:
mode:
Diffstat (limited to 'src/fftw/fni_4.c')
-rw-r--r--src/fftw/fni_4.c102
1 files changed, 102 insertions, 0 deletions
diff --git a/src/fftw/fni_4.c b/src/fftw/fni_4.c
new file mode 100644
index 0000000..f20547e
--- /dev/null
+++ b/src/fftw/fni_4.c
@@ -0,0 +1,102 @@
+/*
+ * Copyright (c) 1997-1999, 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 Mon Mar 24 02:06:18 EST 2003 */
+
+#include "fftw-int.h"
+#include "fftw.h"
+
+/* Generated by: /homee/stevenj/cvs/fftw/gensrc/genfft -magic-alignment-check -magic-twiddle-load-all -magic-variables 4 -magic-loopi -notwiddleinv 4 */
+
+/*
+ * This function contains 16 FP additions, 0 FP multiplications,
+ * (or, 16 additions, 0 multiplications, 0 fused multiply/add),
+ * 12 stack variables, and 16 memory accesses
+ */
+
+/*
+ * Generator Id's :
+ * $Id: fni_4.c,v 1.1 2008/10/17 06:13:18 scuri Exp $
+ * $Id: fni_4.c,v 1.1 2008/10/17 06:13:18 scuri Exp $
+ * $Id: fni_4.c,v 1.1 2008/10/17 06:13:18 scuri Exp $
+ */
+
+void fftwi_no_twiddle_4(const fftw_complex *input, fftw_complex *output,
+ int istride, int ostride)
+{
+ fftw_real tmp3;
+ fftw_real tmp11;
+ fftw_real tmp10;
+ fftw_real tmp15;
+ fftw_real tmp6;
+ fftw_real tmp7;
+ fftw_real tmp14;
+ fftw_real tmp16;
+ ASSERT_ALIGNED_DOUBLE;
+ {
+ fftw_real tmp1;
+ fftw_real tmp2;
+ fftw_real tmp8;
+ fftw_real tmp9;
+ ASSERT_ALIGNED_DOUBLE;
+ tmp1 = c_re(input[0]);
+ tmp2 = c_re(input[2 * istride]);
+ tmp3 = tmp1 + tmp2;
+ tmp11 = tmp1 - tmp2;
+ tmp8 = c_im(input[0]);
+ tmp9 = c_im(input[2 * istride]);
+ tmp10 = tmp8 - tmp9;
+ tmp15 = tmp8 + tmp9;
+ }
+ {
+ fftw_real tmp4;
+ fftw_real tmp5;
+ fftw_real tmp12;
+ fftw_real tmp13;
+ ASSERT_ALIGNED_DOUBLE;
+ tmp4 = c_re(input[istride]);
+ tmp5 = c_re(input[3 * istride]);
+ tmp6 = tmp4 + tmp5;
+ tmp7 = tmp4 - tmp5;
+ tmp12 = c_im(input[istride]);
+ tmp13 = c_im(input[3 * istride]);
+ tmp14 = tmp12 - tmp13;
+ tmp16 = tmp12 + tmp13;
+ }
+ c_re(output[2 * ostride]) = tmp3 - tmp6;
+ c_re(output[0]) = tmp3 + tmp6;
+ c_im(output[ostride]) = tmp7 + tmp10;
+ c_im(output[3 * ostride]) = tmp10 - tmp7;
+ c_re(output[ostride]) = tmp11 - tmp14;
+ c_re(output[3 * ostride]) = tmp11 + tmp14;
+ c_im(output[2 * ostride]) = tmp15 - tmp16;
+ c_im(output[0]) = tmp15 + tmp16;
+}
+
+fftw_codelet_desc fftwi_no_twiddle_4_desc = {
+ "fftwi_no_twiddle_4",
+ (void (*)()) fftwi_no_twiddle_4,
+ 4,
+ FFTW_BACKWARD,
+ FFTW_NOTW,
+ 100,
+ 0,
+ (const int *) 0,
+};