lavu/tx: add support for double precision FFT and MDCT

Simply moves and templates the actual transforms to support an
additional data type.
Unlike the float version, which is equal or better than libfftw3f,
double precision output is bit identical with libfftw3.

doc/APIchanges

@@ -15,6 +15,9 @@ libavutil:     2017-10-21
 
 API changes, most recent first:
 
+2019-07-27 - xxxxxxxxxx - lavu 56.33.100 - tx.h
+  Add AV_TX_DOUBLE_FFT and AV_TX_DOUBLE_MDCT
+
 -------- 8< --------- FFmpeg 4.2 was cut here -------- 8< ---------
 
 2019-06-21 - a30e44098a - lavu 56.30.100 - frame.h

libavutil/Makefile

@@ -161,6 +161,8 @@ OBJS = adler32.o                                                        \
        xtea.o                                                           \
        tea.o                                                            \
        tx.o                                                             \
+       tx_float.o                                                       \
+       tx_double.o
 
 OBJS-$(CONFIG_CUDA)                     += hwcontext_cuda.o
 OBJS-$(CONFIG_D3D11VA)                  += hwcontext_d3d11va.o

libavutil/tx.h

@@ -28,6 +28,10 @@ typedef struct AVComplexFloat {
     float re, im;
 } AVComplexFloat;
 
+typedef struct AVComplexDouble {
+    double re, im;
+} AVComplexDouble;
+
 enum AVTXType {
     /**
      * Standard complex to complex FFT with sample data type AVComplexFloat.
@@ -39,6 +43,14 @@ enum AVTXType {
      * float. Length is the frame size, not the window size (which is 2x frame)
      */
     AV_TX_FLOAT_MDCT = 1,
+    /**
+     * Same as AV_TX_FLOAT_FFT with a data type of AVComplexDouble.
+     */
+    AV_TX_DOUBLE_FFT = 2,
+    /**
+     * Same as AV_TX_FLOAT_MDCT with data and scale type of double.
+     */
+    AV_TX_DOUBLE_MDCT = 3,
 };
 
 /**

libavutil/tx_double.c

+/*
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg 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
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#define TX_DOUBLE
+#include "tx_priv.h"
+#include "tx_template.c"

libavutil/tx_float.c

+/*
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg 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
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#define TX_FLOAT
+#include "tx_priv.h"
+#include "tx_template.c"

libavutil/tx_priv.h

+/*
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg 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
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#ifndef AVUTIL_TX_PRIV_H
+#define AVUTIL_TX_PRIV_H
+
+#include "tx.h"
+#include <stddef.h>
+#include "thread.h"
+#include "mem.h"
+#include "avassert.h"
+#include "attributes.h"
+
+#ifdef TX_FLOAT
+#define TX_NAME(x) x ## _float
+typedef float FFTSample;
+typedef AVComplexFloat FFTComplex;
+#elif defined(TX_DOUBLE)
+#define TX_NAME(x) x ## _double
+typedef double FFTSample;
+typedef AVComplexDouble FFTComplex;
+#else
+typedef void FFTComplex;
+#endif
+
+#if defined(TX_FLOAT) || defined(TX_DOUBLE)
+#define BF(x, y, a, b) do {                                                    \
+        x = (a) - (b);                                                         \
+        y = (a) + (b);                                                         \
+    } while (0)
+
+#define CMUL(dre, dim, are, aim, bre, bim) do {                                \
+        (dre) = (are) * (bre) - (aim) * (bim);                                 \
+        (dim) = (are) * (bim) + (aim) * (bre);                                 \
+    } while (0)
+#endif
+
+#define CMUL3(c, a, b)                                                         \
+    CMUL((c).re, (c).im, (a).re, (a).im, (b).re, (b).im)
+
+#define COSTABLE(size) \
+    DECLARE_ALIGNED(32, FFTSample, TX_NAME(ff_cos_##size))[size/2]
+
+/* Used by asm, reorder with care */
+struct AVTXContext {
+    int n;              /* Nptwo part */
+    int m;              /* Ptwo part */
+    int inv;            /* Is inverted */
+    int type;           /* Type */
+
+    FFTComplex *exptab; /* MDCT exptab */
+    FFTComplex *tmp;    /* Temporary buffer needed for all compound transforms */
+    int        *pfatab; /* Input/Output mapping for compound transforms */
+    int        *revtab; /* Input mapping for power of two transforms */
+};
+
+/* Shared functions */
+int ff_tx_gen_compound_mapping(AVTXContext *s);
+int ff_tx_gen_ptwo_revtab(AVTXContext *s);
+
+/* Also used by SIMD init */
+static inline int split_radix_permutation(int i, int n, int inverse)
+{
+    int m;
+    if (n <= 2)
+        return i & 1;
+    m = n >> 1;
+    if (!(i & m))
+        return split_radix_permutation(i, m, inverse)*2;
+    m >>= 1;
+    if (inverse == !(i & m))
+        return split_radix_permutation(i, m, inverse)*4 + 1;
+    else
+        return split_radix_permutation(i, m, inverse)*4 - 1;
+}
+
+/* Templated functions */
+int ff_tx_init_mdct_fft_float(AVTXContext *s, av_tx_fn *tx,
+                              enum AVTXType type, int inv, int len,
+                              const void *scale, uint64_t flags);
+int ff_tx_init_mdct_fft_double(AVTXContext *s, av_tx_fn *tx,
+                               enum AVTXType type, int inv, int len,
+                               const void *scale, uint64_t flags);
+
+typedef struct CosTabsInitOnce {
+    void (*func)(void);
+    AVOnce control;
+} CosTabsInitOnce;
+
+#endif /* AVUTIL_TX_PRIV_H */

libavutil/version.h

@@ -79,7 +79,7 @@
  */
 
 #define LIBAVUTIL_VERSION_MAJOR  56
-#define LIBAVUTIL_VERSION_MINOR  32
+#define LIBAVUTIL_VERSION_MINOR  33
 #define LIBAVUTIL_VERSION_MICRO 100
 
 #define LIBAVUTIL_VERSION_INT   AV_VERSION_INT(LIBAVUTIL_VERSION_MAJOR, \