x86: dsputil: Split inline assembly from init code
authorDiego Biurrun <diego@biurrun.de>
Tue, 7 May 2013 23:45:39 +0000 (01:45 +0200)
committerDiego Biurrun <diego@biurrun.de>
Sun, 12 May 2013 20:28:07 +0000 (22:28 +0200)
Also remove some pointless comments.

libavcodec/x86/Makefile
libavcodec/x86/dsputil_init.c [new file with mode: 0644]
libavcodec/x86/dsputil_mmx.c
libavcodec/x86/dsputil_mmx.h

index f13f3e0..d58ec84 100644 (file)
@@ -31,7 +31,8 @@ OBJS-$(CONFIG_VP6_DECODER)             += x86/vp56dsp_init.o
 OBJS-$(CONFIG_VP8_DECODER)             += x86/vp8dsp_init.o
 OBJS-$(CONFIG_XMM_CLOBBER_TEST)        += x86/w64xmmtest.o
 
-MMX-OBJS-$(CONFIG_DSPUTIL)             += x86/dsputil_mmx.o             \
+MMX-OBJS-$(CONFIG_DSPUTIL)             += x86/dsputil_init.o            \
+                                          x86/dsputil_mmx.o             \
                                           x86/fdct.o                    \
                                           x86/fpel_mmx.o                \
                                           x86/idct_mmx_xvid.o           \
diff --git a/libavcodec/x86/dsputil_init.c b/libavcodec/x86/dsputil_init.c
new file mode 100644 (file)
index 0000000..2122874
--- /dev/null
@@ -0,0 +1,718 @@
+/*
+ * This file is part of Libav.
+ *
+ * Libav 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.
+ *
+ * Libav 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 Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "config.h"
+#include "libavutil/attributes.h"
+#include "libavutil/cpu.h"
+#include "libavutil/x86/asm.h"
+#include "libavcodec/dsputil.h"
+#include "libavcodec/simple_idct.h"
+#include "dsputil_mmx.h"
+#include "idct_xvid.h"
+
+void ff_put_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2,
+                              int dstStride, int src1Stride, int h);
+void ff_put_no_rnd_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1,
+                                     uint8_t *src2, int dstStride,
+                                     int src1Stride, int h);
+void ff_avg_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2,
+                              int dstStride, int src1Stride, int h);
+void ff_put_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2,
+                               int dstStride, int src1Stride, int h);
+void ff_avg_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2,
+                               int dstStride, int src1Stride, int h);
+void ff_put_no_rnd_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2,
+                                      int dstStride, int src1Stride, int h);
+void ff_put_mpeg4_qpel16_h_lowpass_mmxext(uint8_t *dst, uint8_t *src,
+                                         int dstStride, int srcStride, int h);
+void ff_avg_mpeg4_qpel16_h_lowpass_mmxext(uint8_t *dst, uint8_t *src,
+                                         int dstStride, int srcStride, int h);
+void ff_put_no_rnd_mpeg4_qpel16_h_lowpass_mmxext(uint8_t *dst, uint8_t *src,
+                                                 int dstStride, int srcStride,
+                                                 int h);
+void ff_put_mpeg4_qpel8_h_lowpass_mmxext(uint8_t *dst, uint8_t *src,
+                                        int dstStride, int srcStride, int h);
+void ff_avg_mpeg4_qpel8_h_lowpass_mmxext(uint8_t *dst, uint8_t *src,
+                                        int dstStride, int srcStride, int h);
+void ff_put_no_rnd_mpeg4_qpel8_h_lowpass_mmxext(uint8_t *dst, uint8_t *src,
+                                                int dstStride, int srcStride,
+                                                int h);
+void ff_put_mpeg4_qpel16_v_lowpass_mmxext(uint8_t *dst, uint8_t *src,
+                                         int dstStride, int srcStride);
+void ff_avg_mpeg4_qpel16_v_lowpass_mmxext(uint8_t *dst, uint8_t *src,
+                                         int dstStride, int srcStride);
+void ff_put_no_rnd_mpeg4_qpel16_v_lowpass_mmxext(uint8_t *dst, uint8_t *src,
+                                                 int dstStride, int srcStride);
+void ff_put_mpeg4_qpel8_v_lowpass_mmxext(uint8_t *dst, uint8_t *src,
+                                        int dstStride, int srcStride);
+void ff_avg_mpeg4_qpel8_v_lowpass_mmxext(uint8_t *dst, uint8_t *src,
+                                        int dstStride, int srcStride);
+void ff_put_no_rnd_mpeg4_qpel8_v_lowpass_mmxext(uint8_t *dst, uint8_t *src,
+                                                int dstStride, int srcStride);
+#define ff_put_no_rnd_pixels16_mmxext ff_put_pixels16_mmxext
+#define ff_put_no_rnd_pixels8_mmxext ff_put_pixels8_mmxext
+
+void ff_h263_v_loop_filter_mmx(uint8_t *src, int stride, int qscale);
+void ff_h263_h_loop_filter_mmx(uint8_t *src, int stride, int qscale);
+
+int32_t ff_scalarproduct_int16_mmxext(const int16_t *v1, const int16_t *v2,
+                                      int order);
+int32_t ff_scalarproduct_int16_sse2(const int16_t *v1, const int16_t *v2,
+                                    int order);
+int32_t ff_scalarproduct_and_madd_int16_mmxext(int16_t *v1, const int16_t *v2,
+                                               const int16_t *v3,
+                                               int order, int mul);
+int32_t ff_scalarproduct_and_madd_int16_sse2(int16_t *v1, const int16_t *v2,
+                                             const int16_t *v3,
+                                             int order, int mul);
+int32_t ff_scalarproduct_and_madd_int16_ssse3(int16_t *v1, const int16_t *v2,
+                                              const int16_t *v3,
+                                              int order, int mul);
+
+void ff_apply_window_int16_round_mmxext(int16_t *output, const int16_t *input,
+                                        const int16_t *window, unsigned int len);
+void ff_apply_window_int16_round_sse2(int16_t *output, const int16_t *input,
+                                      const int16_t *window, unsigned int len);
+void ff_apply_window_int16_mmxext(int16_t *output, const int16_t *input,
+                                  const int16_t *window, unsigned int len);
+void ff_apply_window_int16_sse2(int16_t *output, const int16_t *input,
+                                const int16_t *window, unsigned int len);
+void ff_apply_window_int16_ssse3(int16_t *output, const int16_t *input,
+                                 const int16_t *window, unsigned int len);
+void ff_apply_window_int16_ssse3_atom(int16_t *output, const int16_t *input,
+                                      const int16_t *window, unsigned int len);
+
+void ff_bswap32_buf_ssse3(uint32_t *dst, const uint32_t *src, int w);
+void ff_bswap32_buf_sse2(uint32_t *dst, const uint32_t *src, int w);
+
+void ff_add_hfyu_median_prediction_mmxext(uint8_t *dst, const uint8_t *top,
+                                          const uint8_t *diff, int w,
+                                          int *left, int *left_top);
+int  ff_add_hfyu_left_prediction_ssse3(uint8_t *dst, const uint8_t *src,
+                                       int w, int left);
+int  ff_add_hfyu_left_prediction_sse4(uint8_t *dst, const uint8_t *src,
+                                      int w, int left);
+
+void ff_vector_clip_int32_mmx     (int32_t *dst, const int32_t *src,
+                                   int32_t min, int32_t max, unsigned int len);
+void ff_vector_clip_int32_sse2    (int32_t *dst, const int32_t *src,
+                                   int32_t min, int32_t max, unsigned int len);
+void ff_vector_clip_int32_int_sse2(int32_t *dst, const int32_t *src,
+                                   int32_t min, int32_t max, unsigned int len);
+void ff_vector_clip_int32_sse4    (int32_t *dst, const int32_t *src,
+                                   int32_t min, int32_t max, unsigned int len);
+
+#if HAVE_YASM
+
+PIXELS16(static, ff_avg, , , _mmxext)
+PIXELS16(static, ff_put, , , _mmxext)
+
+#define QPEL_OP(OPNAME, RND, MMX)                                       \
+static void OPNAME ## qpel8_mc00_ ## MMX (uint8_t *dst, uint8_t *src,   \
+                                          ptrdiff_t stride)             \
+{                                                                       \
+    ff_ ## OPNAME ## pixels8_ ## MMX(dst, src, stride, 8);              \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel8_mc10_ ## MMX(uint8_t *dst, uint8_t *src,    \
+                                         ptrdiff_t stride)              \
+{                                                                       \
+    uint64_t temp[8];                                                   \
+    uint8_t * const half = (uint8_t*)temp;                              \
+    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(half, src, 8,        \
+                                                   stride, 8);          \
+    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src, half,                 \
+                                        stride, stride, 8);             \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel8_mc20_ ## MMX(uint8_t *dst, uint8_t *src,    \
+                                         ptrdiff_t stride)              \
+{                                                                       \
+    ff_ ## OPNAME ## mpeg4_qpel8_h_lowpass_ ## MMX(dst, src, stride,    \
+                                                   stride, 8);          \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel8_mc30_ ## MMX(uint8_t *dst, uint8_t *src,    \
+                                         ptrdiff_t stride)              \
+{                                                                       \
+    uint64_t temp[8];                                                   \
+    uint8_t * const half = (uint8_t*)temp;                              \
+    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(half, src, 8,        \
+                                                   stride, 8);          \
+    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src + 1, half, stride,     \
+                                        stride, 8);                     \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel8_mc01_ ## MMX(uint8_t *dst, uint8_t *src,    \
+                                         ptrdiff_t stride)              \
+{                                                                       \
+    uint64_t temp[8];                                                   \
+    uint8_t * const half = (uint8_t*)temp;                              \
+    ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(half, src,           \
+                                                   8, stride);          \
+    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src, half,                 \
+                                        stride, stride, 8);             \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel8_mc02_ ## MMX(uint8_t *dst, uint8_t *src,    \
+                                         ptrdiff_t stride)              \
+{                                                                       \
+    ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, src,            \
+                                                   stride, stride);     \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel8_mc03_ ## MMX(uint8_t *dst, uint8_t *src,    \
+                                         ptrdiff_t stride)              \
+{                                                                       \
+    uint64_t temp[8];                                                   \
+    uint8_t * const half = (uint8_t*)temp;                              \
+    ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(half, src,           \
+                                                   8, stride);          \
+    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src + stride, half, stride,\
+                                        stride, 8);                     \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel8_mc11_ ## MMX(uint8_t *dst, uint8_t *src,    \
+                                         ptrdiff_t stride)              \
+{                                                                       \
+    uint64_t half[8 + 9];                                               \
+    uint8_t * const halfH  = ((uint8_t*)half) + 64;                     \
+    uint8_t * const halfHV = ((uint8_t*)half);                          \
+    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
+                                                   stride, 9);          \
+    ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH, 8,           \
+                                        stride, 9);                     \
+    ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
+    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV,             \
+                                        stride, 8, 8);                  \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel8_mc31_ ## MMX(uint8_t *dst, uint8_t *src,    \
+                                         ptrdiff_t stride)              \
+{                                                                       \
+    uint64_t half[8 + 9];                                               \
+    uint8_t * const halfH  = ((uint8_t*)half) + 64;                     \
+    uint8_t * const halfHV = ((uint8_t*)half);                          \
+    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
+                                                   stride, 9);          \
+    ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src + 1, halfH, 8,       \
+                                        stride, 9);                     \
+    ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
+    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV,             \
+                                        stride, 8, 8);                  \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel8_mc13_ ## MMX(uint8_t *dst, uint8_t *src,    \
+                                         ptrdiff_t stride)              \
+{                                                                       \
+    uint64_t half[8 + 9];                                               \
+    uint8_t * const halfH  = ((uint8_t*)half) + 64;                     \
+    uint8_t * const halfHV = ((uint8_t*)half);                          \
+    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
+                                                   stride, 9);          \
+    ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH, 8,           \
+                                        stride, 9);                     \
+    ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
+    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH + 8, halfHV,         \
+                                        stride, 8, 8);                  \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel8_mc33_ ## MMX(uint8_t *dst, uint8_t *src,    \
+                                         ptrdiff_t stride)              \
+{                                                                       \
+    uint64_t half[8 + 9];                                               \
+    uint8_t * const halfH  = ((uint8_t*)half) + 64;                     \
+    uint8_t * const halfHV = ((uint8_t*)half);                          \
+    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
+                                                   stride, 9);          \
+    ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src + 1, halfH, 8,       \
+                                        stride, 9);                     \
+    ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
+    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH + 8, halfHV,         \
+                                        stride, 8, 8);                  \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel8_mc21_ ## MMX(uint8_t *dst, uint8_t *src,    \
+                                         ptrdiff_t stride)              \
+{                                                                       \
+    uint64_t half[8 + 9];                                               \
+    uint8_t * const halfH  = ((uint8_t*)half) + 64;                     \
+    uint8_t * const halfHV = ((uint8_t*)half);                          \
+    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
+                                                   stride, 9);          \
+    ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
+    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV,             \
+                                        stride, 8, 8);                  \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel8_mc23_ ## MMX(uint8_t *dst, uint8_t *src,    \
+                                         ptrdiff_t stride)              \
+{                                                                       \
+    uint64_t half[8 + 9];                                               \
+    uint8_t * const halfH  = ((uint8_t*)half) + 64;                     \
+    uint8_t * const halfHV = ((uint8_t*)half);                          \
+    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
+                                                   stride, 9);          \
+    ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
+    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH + 8, halfHV,         \
+                                        stride, 8, 8);                  \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel8_mc12_ ## MMX(uint8_t *dst, uint8_t *src,    \
+                                         ptrdiff_t stride)              \
+{                                                                       \
+    uint64_t half[8 + 9];                                               \
+    uint8_t * const halfH = ((uint8_t*)half);                           \
+    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
+                                                   stride, 9);          \
+    ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH,              \
+                                        8, stride, 9);                  \
+    ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH,          \
+                                                   stride, 8);          \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel8_mc32_ ## MMX(uint8_t *dst, uint8_t *src,    \
+                                         ptrdiff_t stride)              \
+{                                                                       \
+    uint64_t half[8 + 9];                                               \
+    uint8_t * const halfH = ((uint8_t*)half);                           \
+    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
+                                                   stride, 9);          \
+    ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src + 1, halfH, 8,       \
+                                        stride, 9);                     \
+    ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH,          \
+                                                   stride, 8);          \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel8_mc22_ ## MMX(uint8_t *dst, uint8_t *src,    \
+                                         ptrdiff_t stride)              \
+{                                                                       \
+    uint64_t half[9];                                                   \
+    uint8_t * const halfH = ((uint8_t*)half);                           \
+    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
+                                                   stride, 9);          \
+    ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH,          \
+                                                   stride, 8);          \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel16_mc00_ ## MMX (uint8_t *dst, uint8_t *src,  \
+                                           ptrdiff_t stride)            \
+{                                                                       \
+    ff_ ## OPNAME ## pixels16_ ## MMX(dst, src, stride, 16);            \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel16_mc10_ ## MMX(uint8_t *dst, uint8_t *src,   \
+                                          ptrdiff_t stride)             \
+{                                                                       \
+    uint64_t temp[32];                                                  \
+    uint8_t * const half = (uint8_t*)temp;                              \
+    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(half, src, 16,      \
+                                                    stride, 16);        \
+    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src, half, stride,        \
+                                         stride, 16);                   \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel16_mc20_ ## MMX(uint8_t *dst, uint8_t *src,   \
+                                          ptrdiff_t stride)             \
+{                                                                       \
+    ff_ ## OPNAME ## mpeg4_qpel16_h_lowpass_ ## MMX(dst, src,           \
+                                                    stride, stride, 16);\
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel16_mc30_ ## MMX(uint8_t *dst, uint8_t *src,   \
+                                          ptrdiff_t stride)             \
+{                                                                       \
+    uint64_t temp[32];                                                  \
+    uint8_t * const half = (uint8_t*)temp;                              \
+    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(half, src, 16,      \
+                                                    stride, 16);        \
+    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src + 1, half,            \
+                                         stride, stride, 16);           \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel16_mc01_ ## MMX(uint8_t *dst, uint8_t *src,   \
+                                          ptrdiff_t stride)             \
+{                                                                       \
+    uint64_t temp[32];                                                  \
+    uint8_t * const half = (uint8_t*)temp;                              \
+    ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(half, src, 16,      \
+                                                    stride);            \
+    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src, half, stride,        \
+                                         stride, 16);                   \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel16_mc02_ ## MMX(uint8_t *dst, uint8_t *src,   \
+                                          ptrdiff_t stride)             \
+{                                                                       \
+    ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, src,           \
+                                                    stride, stride);    \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel16_mc03_ ## MMX(uint8_t *dst, uint8_t *src,   \
+                                          ptrdiff_t stride)             \
+{                                                                       \
+    uint64_t temp[32];                                                  \
+    uint8_t * const half = (uint8_t*)temp;                              \
+    ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(half, src, 16,      \
+                                                    stride);            \
+    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src+stride, half,         \
+                                         stride, stride, 16);           \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel16_mc11_ ## MMX(uint8_t *dst, uint8_t *src,   \
+                                          ptrdiff_t stride)             \
+{                                                                       \
+    uint64_t half[16 * 2 + 17 * 2];                                     \
+    uint8_t * const halfH  = ((uint8_t*)half) + 256;                    \
+    uint8_t * const halfHV = ((uint8_t*)half);                          \
+    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
+                                                    stride, 17);        \
+    ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16,         \
+                                         stride, 17);                   \
+    ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH,      \
+                                                    16, 16);            \
+    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV,            \
+                                         stride, 16, 16);               \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel16_mc31_ ## MMX(uint8_t *dst, uint8_t *src,   \
+                                          ptrdiff_t stride)             \
+{                                                                       \
+    uint64_t half[16 * 2 + 17 * 2];                                     \
+    uint8_t * const halfH  = ((uint8_t*)half) + 256;                    \
+    uint8_t * const halfHV = ((uint8_t*)half);                          \
+    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
+                                                    stride, 17);        \
+    ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src + 1, halfH, 16,     \
+                                         stride, 17);                   \
+    ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH,      \
+                                                    16, 16);            \
+    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV,            \
+                                         stride, 16, 16);               \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel16_mc13_ ## MMX(uint8_t *dst, uint8_t *src,   \
+                                          ptrdiff_t stride)             \
+{                                                                       \
+    uint64_t half[16 * 2 + 17 * 2];                                     \
+    uint8_t * const halfH  = ((uint8_t*)half) + 256;                    \
+    uint8_t * const halfHV = ((uint8_t*)half);                          \
+    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
+                                                    stride, 17);        \
+    ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16,         \
+                                         stride, 17);                   \
+    ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH,      \
+                                                    16, 16);            \
+    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH + 16, halfHV,       \
+                                         stride, 16, 16);               \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel16_mc33_ ## MMX(uint8_t *dst, uint8_t *src,   \
+                                          ptrdiff_t stride)             \
+{                                                                       \
+    uint64_t half[16 * 2 + 17 * 2];                                     \
+    uint8_t * const halfH  = ((uint8_t*)half) + 256;                    \
+    uint8_t * const halfHV = ((uint8_t*)half);                          \
+    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
+                                                    stride, 17);        \
+    ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src + 1, halfH, 16,     \
+                                         stride, 17);                   \
+    ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH,      \
+                                                    16, 16);            \
+    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH + 16, halfHV,       \
+                                         stride, 16, 16);               \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel16_mc21_ ## MMX(uint8_t *dst, uint8_t *src,   \
+                                          ptrdiff_t stride)             \
+{                                                                       \
+    uint64_t half[16 * 2 + 17 * 2];                                     \
+    uint8_t * const halfH  = ((uint8_t*)half) + 256;                    \
+    uint8_t * const halfHV = ((uint8_t*)half);                          \
+    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
+                                                    stride, 17);        \
+    ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH,      \
+                                                    16, 16);            \
+    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV,            \
+                                         stride, 16, 16);               \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel16_mc23_ ## MMX(uint8_t *dst, uint8_t *src,   \
+                                          ptrdiff_t stride)             \
+{                                                                       \
+    uint64_t half[16 * 2 + 17 * 2];                                     \
+    uint8_t * const halfH  = ((uint8_t*)half) + 256;                    \
+    uint8_t * const halfHV = ((uint8_t*)half);                          \
+    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
+                                                    stride, 17);        \
+    ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH,      \
+                                                    16, 16);            \
+    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH + 16, halfHV,       \
+                                         stride, 16, 16);               \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel16_mc12_ ## MMX(uint8_t *dst, uint8_t *src,   \
+                                          ptrdiff_t stride)             \
+{                                                                       \
+    uint64_t half[17 * 2];                                              \
+    uint8_t * const halfH = ((uint8_t*)half);                           \
+    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
+                                                    stride, 17);        \
+    ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16,         \
+                                         stride, 17);                   \
+    ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH,         \
+                                                    stride, 16);        \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel16_mc32_ ## MMX(uint8_t *dst, uint8_t *src,   \
+                                          ptrdiff_t stride)             \
+{                                                                       \
+    uint64_t half[17 * 2];                                              \
+    uint8_t * const halfH = ((uint8_t*)half);                           \
+    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
+                                                    stride, 17);        \
+    ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src + 1, halfH, 16,     \
+                                         stride, 17);                   \
+    ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH,         \
+                                                    stride, 16);        \
+}                                                                       \
+                                                                        \
+static void OPNAME ## qpel16_mc22_ ## MMX(uint8_t *dst, uint8_t *src,   \
+                                          ptrdiff_t stride)             \
+{                                                                       \
+    uint64_t half[17 * 2];                                              \
+    uint8_t * const halfH = ((uint8_t*)half);                           \
+    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
+                                                    stride, 17);        \
+    ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH,         \
+                                                    stride, 16);        \
+}
+
+QPEL_OP(put_,        _,        mmxext)
+QPEL_OP(avg_,        _,        mmxext)
+QPEL_OP(put_no_rnd_, _no_rnd_, mmxext)
+#endif /* HAVE_YASM */
+
+#define SET_QPEL_FUNCS(PFX, IDX, SIZE, CPU, PREFIX)                          \
+    do {                                                                     \
+    c->PFX ## _pixels_tab[IDX][ 0] = PREFIX ## PFX ## SIZE ## _mc00_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 1] = PREFIX ## PFX ## SIZE ## _mc10_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 2] = PREFIX ## PFX ## SIZE ## _mc20_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 3] = PREFIX ## PFX ## SIZE ## _mc30_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 4] = PREFIX ## PFX ## SIZE ## _mc01_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 5] = PREFIX ## PFX ## SIZE ## _mc11_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 6] = PREFIX ## PFX ## SIZE ## _mc21_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 7] = PREFIX ## PFX ## SIZE ## _mc31_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 8] = PREFIX ## PFX ## SIZE ## _mc02_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 9] = PREFIX ## PFX ## SIZE ## _mc12_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][10] = PREFIX ## PFX ## SIZE ## _mc22_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][11] = PREFIX ## PFX ## SIZE ## _mc32_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][12] = PREFIX ## PFX ## SIZE ## _mc03_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][13] = PREFIX ## PFX ## SIZE ## _mc13_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][14] = PREFIX ## PFX ## SIZE ## _mc23_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][15] = PREFIX ## PFX ## SIZE ## _mc33_ ## CPU; \
+    } while (0)
+
+static av_cold void dsputil_init_mmx(DSPContext *c, AVCodecContext *avctx,
+                                     int mm_flags)
+{
+#if HAVE_MMX_INLINE
+    const int high_bit_depth = avctx->bits_per_raw_sample > 8;
+
+    c->put_pixels_clamped        = ff_put_pixels_clamped_mmx;
+    c->put_signed_pixels_clamped = ff_put_signed_pixels_clamped_mmx;
+    c->add_pixels_clamped        = ff_add_pixels_clamped_mmx;
+
+    if (!high_bit_depth) {
+        c->clear_block  = ff_clear_block_mmx;
+        c->clear_blocks = ff_clear_blocks_mmx;
+        c->draw_edges   = ff_draw_edges_mmx;
+
+        switch (avctx->idct_algo) {
+        case FF_IDCT_AUTO:
+        case FF_IDCT_SIMPLEMMX:
+            c->idct_put              = ff_simple_idct_put_mmx;
+            c->idct_add              = ff_simple_idct_add_mmx;
+            c->idct                  = ff_simple_idct_mmx;
+            c->idct_permutation_type = FF_SIMPLE_IDCT_PERM;
+            break;
+        case FF_IDCT_XVIDMMX:
+            c->idct_put              = ff_idct_xvid_mmx_put;
+            c->idct_add              = ff_idct_xvid_mmx_add;
+            c->idct                  = ff_idct_xvid_mmx;
+            break;
+        }
+    }
+
+    c->gmc = ff_gmc_mmx;
+
+    c->add_bytes = ff_add_bytes_mmx;
+#endif /* HAVE_MMX_INLINE */
+
+#if HAVE_MMX_EXTERNAL
+    if (CONFIG_H263_DECODER || CONFIG_H263_ENCODER) {
+        c->h263_v_loop_filter = ff_h263_v_loop_filter_mmx;
+        c->h263_h_loop_filter = ff_h263_h_loop_filter_mmx;
+    }
+
+    c->vector_clip_int32 = ff_vector_clip_int32_mmx;
+#endif /* HAVE_MMX_EXTERNAL */
+}
+
+static av_cold void dsputil_init_mmxext(DSPContext *c, AVCodecContext *avctx,
+                                        int mm_flags)
+{
+#if HAVE_MMXEXT_INLINE
+    const int high_bit_depth = avctx->bits_per_raw_sample > 8;
+
+    if (!high_bit_depth && avctx->idct_algo == FF_IDCT_XVIDMMX) {
+        c->idct_put = ff_idct_xvid_mmxext_put;
+        c->idct_add = ff_idct_xvid_mmxext_add;
+        c->idct     = ff_idct_xvid_mmxext;
+    }
+#endif /* HAVE_MMXEXT_INLINE */
+
+#if HAVE_MMXEXT_EXTERNAL
+    SET_QPEL_FUNCS(avg_qpel,        0, 16, mmxext, );
+    SET_QPEL_FUNCS(avg_qpel,        1,  8, mmxext, );
+
+    SET_QPEL_FUNCS(put_qpel,        0, 16, mmxext, );
+    SET_QPEL_FUNCS(put_qpel,        1,  8, mmxext, );
+    SET_QPEL_FUNCS(put_no_rnd_qpel, 0, 16, mmxext, );
+    SET_QPEL_FUNCS(put_no_rnd_qpel, 1,  8, mmxext, );
+
+    /* slower than cmov version on AMD */
+    if (!(mm_flags & AV_CPU_FLAG_3DNOW))
+        c->add_hfyu_median_prediction = ff_add_hfyu_median_prediction_mmxext;
+
+    c->scalarproduct_int16          = ff_scalarproduct_int16_mmxext;
+    c->scalarproduct_and_madd_int16 = ff_scalarproduct_and_madd_int16_mmxext;
+
+    if (avctx->flags & CODEC_FLAG_BITEXACT) {
+        c->apply_window_int16 = ff_apply_window_int16_mmxext;
+    } else {
+        c->apply_window_int16 = ff_apply_window_int16_round_mmxext;
+    }
+#endif /* HAVE_MMXEXT_EXTERNAL */
+}
+
+static av_cold void dsputil_init_sse(DSPContext *c, AVCodecContext *avctx,
+                                     int mm_flags)
+{
+#if HAVE_SSE_INLINE
+    const int high_bit_depth = avctx->bits_per_raw_sample > 8;
+
+    if (!high_bit_depth) {
+        if (!(CONFIG_MPEG_XVMC_DECODER && avctx->xvmc_acceleration > 1)) {
+            /* XvMCCreateBlocks() may not allocate 16-byte aligned blocks */
+            c->clear_block  = ff_clear_block_sse;
+            c->clear_blocks = ff_clear_blocks_sse;
+        }
+    }
+
+    c->vector_clipf = ff_vector_clipf_sse;
+#endif /* HAVE_SSE_INLINE */
+}
+
+static av_cold void dsputil_init_sse2(DSPContext *c, AVCodecContext *avctx,
+                                      int mm_flags)
+{
+#if HAVE_SSE2_INLINE
+    const int high_bit_depth = avctx->bits_per_raw_sample > 8;
+
+    if (!high_bit_depth && avctx->idct_algo == FF_IDCT_XVIDMMX) {
+        c->idct_put              = ff_idct_xvid_sse2_put;
+        c->idct_add              = ff_idct_xvid_sse2_add;
+        c->idct                  = ff_idct_xvid_sse2;
+        c->idct_permutation_type = FF_SSE2_IDCT_PERM;
+    }
+#endif /* HAVE_SSE2_INLINE */
+
+#if HAVE_SSE2_EXTERNAL
+    c->scalarproduct_int16          = ff_scalarproduct_int16_sse2;
+    c->scalarproduct_and_madd_int16 = ff_scalarproduct_and_madd_int16_sse2;
+    if (mm_flags & AV_CPU_FLAG_ATOM) {
+        c->vector_clip_int32 = ff_vector_clip_int32_int_sse2;
+    } else {
+        c->vector_clip_int32 = ff_vector_clip_int32_sse2;
+    }
+    if (avctx->flags & CODEC_FLAG_BITEXACT) {
+        c->apply_window_int16 = ff_apply_window_int16_sse2;
+    } else if (!(mm_flags & AV_CPU_FLAG_SSE2SLOW)) {
+        c->apply_window_int16 = ff_apply_window_int16_round_sse2;
+    }
+    c->bswap_buf = ff_bswap32_buf_sse2;
+#endif /* HAVE_SSE2_EXTERNAL */
+}
+
+static av_cold void dsputil_init_ssse3(DSPContext *c, AVCodecContext *avctx,
+                                       int mm_flags)
+{
+#if HAVE_SSSE3_EXTERNAL
+    c->add_hfyu_left_prediction = ff_add_hfyu_left_prediction_ssse3;
+    if (mm_flags & AV_CPU_FLAG_SSE4) // not really sse4, just slow on Conroe
+        c->add_hfyu_left_prediction = ff_add_hfyu_left_prediction_sse4;
+
+    if (mm_flags & AV_CPU_FLAG_ATOM)
+        c->apply_window_int16 = ff_apply_window_int16_ssse3_atom;
+    else
+        c->apply_window_int16 = ff_apply_window_int16_ssse3;
+    if (!(mm_flags & (AV_CPU_FLAG_SSE42|AV_CPU_FLAG_3DNOW))) // cachesplit
+        c->scalarproduct_and_madd_int16 = ff_scalarproduct_and_madd_int16_ssse3;
+    c->bswap_buf = ff_bswap32_buf_ssse3;
+#endif /* HAVE_SSSE3_EXTERNAL */
+}
+
+static av_cold void dsputil_init_sse4(DSPContext *c, AVCodecContext *avctx,
+                                      int mm_flags)
+{
+#if HAVE_SSE4_EXTERNAL
+    c->vector_clip_int32 = ff_vector_clip_int32_sse4;
+#endif /* HAVE_SSE4_EXTERNAL */
+}
+
+av_cold void ff_dsputil_init_mmx(DSPContext *c, AVCodecContext *avctx)
+{
+    int mm_flags = av_get_cpu_flags();
+
+#if HAVE_7REGS && HAVE_INLINE_ASM
+    if (mm_flags & AV_CPU_FLAG_CMOV)
+        c->add_hfyu_median_prediction = ff_add_hfyu_median_prediction_cmov;
+#endif
+
+    if (mm_flags & AV_CPU_FLAG_MMX)
+        dsputil_init_mmx(c, avctx, mm_flags);
+
+    if (mm_flags & AV_CPU_FLAG_MMXEXT)
+        dsputil_init_mmxext(c, avctx, mm_flags);
+
+    if (mm_flags & AV_CPU_FLAG_SSE)
+        dsputil_init_sse(c, avctx, mm_flags);
+
+    if (mm_flags & AV_CPU_FLAG_SSE2)
+        dsputil_init_sse2(c, avctx, mm_flags);
+
+    if (mm_flags & AV_CPU_FLAG_SSSE3)
+        dsputil_init_ssse3(c, avctx, mm_flags);
+
+    if (mm_flags & AV_CPU_FLAG_SSE4)
+        dsputil_init_sse4(c, avctx, mm_flags);
+
+    if (CONFIG_ENCODERS)
+        ff_dsputilenc_init_mmx(c, avctx);
+}
index 89da178..832ee5a 100644 (file)
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#include "libavutil/attributes.h"
+#include "config.h"
 #include "libavutil/cpu.h"
 #include "libavutil/x86/asm.h"
-#include "libavcodec/dsputil.h"
-#include "libavcodec/h264dsp.h"
-#include "libavcodec/mpegvideo.h"
-#include "libavcodec/simple_idct.h"
 #include "constants.h"
 #include "dsputil_mmx.h"
-#include "idct_xvid.h"
-
-//#undef NDEBUG
-//#include <assert.h>
-
-void ff_put_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2,
-                              int dstStride, int src1Stride, int h);
-void ff_put_no_rnd_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1,
-                                     uint8_t *src2, int dstStride,
-                                     int src1Stride, int h);
-void ff_avg_pixels8_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2,
-                              int dstStride, int src1Stride, int h);
-void ff_put_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2,
-                               int dstStride, int src1Stride, int h);
-void ff_avg_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2,
-                               int dstStride, int src1Stride, int h);
-void ff_put_no_rnd_pixels16_l2_mmxext(uint8_t *dst, uint8_t *src1, uint8_t *src2,
-                                      int dstStride, int src1Stride, int h);
-void ff_put_mpeg4_qpel16_h_lowpass_mmxext(uint8_t *dst, uint8_t *src,
-                                         int dstStride, int srcStride, int h);
-void ff_avg_mpeg4_qpel16_h_lowpass_mmxext(uint8_t *dst, uint8_t *src,
-                                         int dstStride, int srcStride, int h);
-void ff_put_no_rnd_mpeg4_qpel16_h_lowpass_mmxext(uint8_t *dst, uint8_t *src,
-                                                 int dstStride, int srcStride,
-                                                 int h);
-void ff_put_mpeg4_qpel8_h_lowpass_mmxext(uint8_t *dst, uint8_t *src,
-                                        int dstStride, int srcStride, int h);
-void ff_avg_mpeg4_qpel8_h_lowpass_mmxext(uint8_t *dst, uint8_t *src,
-                                        int dstStride, int srcStride, int h);
-void ff_put_no_rnd_mpeg4_qpel8_h_lowpass_mmxext(uint8_t *dst, uint8_t *src,
-                                                int dstStride, int srcStride,
-                                                int h);
-void ff_put_mpeg4_qpel16_v_lowpass_mmxext(uint8_t *dst, uint8_t *src,
-                                         int dstStride, int srcStride);
-void ff_avg_mpeg4_qpel16_v_lowpass_mmxext(uint8_t *dst, uint8_t *src,
-                                         int dstStride, int srcStride);
-void ff_put_no_rnd_mpeg4_qpel16_v_lowpass_mmxext(uint8_t *dst, uint8_t *src,
-                                                 int dstStride, int srcStride);
-void ff_put_mpeg4_qpel8_v_lowpass_mmxext(uint8_t *dst, uint8_t *src,
-                                        int dstStride, int srcStride);
-void ff_avg_mpeg4_qpel8_v_lowpass_mmxext(uint8_t *dst, uint8_t *src,
-                                        int dstStride, int srcStride);
-void ff_put_no_rnd_mpeg4_qpel8_v_lowpass_mmxext(uint8_t *dst, uint8_t *src,
-                                                int dstStride, int srcStride);
-#define ff_put_no_rnd_pixels16_mmxext ff_put_pixels16_mmxext
-#define ff_put_no_rnd_pixels8_mmxext ff_put_pixels8_mmxext
-
 
 #if HAVE_INLINE_ASM
 
-/***********************************/
-/* standard MMX */
-
 void ff_put_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels,
                                int line_size)
 {
@@ -219,7 +165,7 @@ void ff_add_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels,
 }
 
 #define CLEAR_BLOCKS(name, n)                           \
-static void name(int16_t *blocks)                       \
+void name(int16_t *blocks)                              \
 {                                                       \
     __asm__ volatile (                                  \
         "pxor %%mm7, %%mm7              \n\t"           \
@@ -236,10 +182,10 @@ static void name(int16_t *blocks)                       \
         : "%"REG_a                                      \
         );                                              \
 }
-CLEAR_BLOCKS(clear_blocks_mmx, 6)
-CLEAR_BLOCKS(clear_block_mmx, 1)
+CLEAR_BLOCKS(ff_clear_blocks_mmx, 6)
+CLEAR_BLOCKS(ff_clear_block_mmx, 1)
 
-static void clear_block_sse(int16_t *block)
+void ff_clear_block_sse(int16_t *block)
 {
     __asm__ volatile (
         "xorps  %%xmm0, %%xmm0          \n"
@@ -256,7 +202,7 @@ static void clear_block_sse(int16_t *block)
     );
 }
 
-static void clear_blocks_sse(int16_t *blocks)
+void ff_clear_blocks_sse(int16_t *blocks)
 {
     __asm__ volatile (
         "xorps  %%xmm0, %%xmm0              \n"
@@ -278,7 +224,7 @@ static void clear_blocks_sse(int16_t *blocks)
     );
 }
 
-static void add_bytes_mmx(uint8_t *dst, uint8_t *src, int w)
+void ff_add_bytes_mmx(uint8_t *dst, uint8_t *src, int w)
 {
     x86_reg i = 0;
     __asm__ volatile (
@@ -304,9 +250,9 @@ static void add_bytes_mmx(uint8_t *dst, uint8_t *src, int w)
 }
 
 #if HAVE_7REGS
-static void add_hfyu_median_prediction_cmov(uint8_t *dst, const uint8_t *top,
-                                            const uint8_t *diff, int w,
-                                            int *left, int *left_top)
+void ff_add_hfyu_median_prediction_cmov(uint8_t *dst, const uint8_t *top,
+                                        const uint8_t *diff, int w,
+                                        int *left, int *left_top)
 {
     x86_reg w2 = -w;
     x86_reg x;
@@ -343,8 +289,8 @@ static void add_hfyu_median_prediction_cmov(uint8_t *dst, const uint8_t *top,
 
 /* Draw the edges of width 'w' of an image of size width, height
  * this MMX version can only handle w == 8 || w == 16. */
-static void draw_edges_mmx(uint8_t *buf, int wrap, int width, int height,
-                           int w, int h, int sides)
+void ff_draw_edges_mmx(uint8_t *buf, int wrap, int width, int height,
+                       int w, int h, int sides)
 {
     uint8_t *ptr, *last_line;
     int i;
@@ -436,406 +382,11 @@ static void draw_edges_mmx(uint8_t *buf, int wrap, int width, int height,
         }
     }
 }
-#endif /* HAVE_INLINE_ASM */
-
-
-#if HAVE_YASM
-
-PIXELS16(static, ff_avg, , , _mmxext)
-PIXELS16(static, ff_put, , , _mmxext)
-
-#define QPEL_OP(OPNAME, RND, MMX)                                       \
-static void OPNAME ## qpel8_mc00_ ## MMX (uint8_t *dst, uint8_t *src,   \
-                                          ptrdiff_t stride)             \
-{                                                                       \
-    ff_ ## OPNAME ## pixels8_ ## MMX(dst, src, stride, 8);              \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel8_mc10_ ## MMX(uint8_t *dst, uint8_t *src,    \
-                                         ptrdiff_t stride)              \
-{                                                                       \
-    uint64_t temp[8];                                                   \
-    uint8_t * const half = (uint8_t*)temp;                              \
-    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(half, src, 8,        \
-                                                   stride, 8);          \
-    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src, half,                 \
-                                        stride, stride, 8);             \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel8_mc20_ ## MMX(uint8_t *dst, uint8_t *src,    \
-                                         ptrdiff_t stride)              \
-{                                                                       \
-    ff_ ## OPNAME ## mpeg4_qpel8_h_lowpass_ ## MMX(dst, src, stride,    \
-                                                   stride, 8);          \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel8_mc30_ ## MMX(uint8_t *dst, uint8_t *src,    \
-                                         ptrdiff_t stride)              \
-{                                                                       \
-    uint64_t temp[8];                                                   \
-    uint8_t * const half = (uint8_t*)temp;                              \
-    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(half, src, 8,        \
-                                                   stride, 8);          \
-    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src + 1, half, stride,     \
-                                        stride, 8);                     \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel8_mc01_ ## MMX(uint8_t *dst, uint8_t *src,    \
-                                         ptrdiff_t stride)              \
-{                                                                       \
-    uint64_t temp[8];                                                   \
-    uint8_t * const half = (uint8_t*)temp;                              \
-    ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(half, src,           \
-                                                   8, stride);          \
-    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src, half,                 \
-                                        stride, stride, 8);             \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel8_mc02_ ## MMX(uint8_t *dst, uint8_t *src,    \
-                                         ptrdiff_t stride)              \
-{                                                                       \
-    ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, src,            \
-                                                   stride, stride);     \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel8_mc03_ ## MMX(uint8_t *dst, uint8_t *src,    \
-                                         ptrdiff_t stride)              \
-{                                                                       \
-    uint64_t temp[8];                                                   \
-    uint8_t * const half = (uint8_t*)temp;                              \
-    ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(half, src,           \
-                                                   8, stride);          \
-    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, src + stride, half, stride,\
-                                        stride, 8);                     \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel8_mc11_ ## MMX(uint8_t *dst, uint8_t *src,    \
-                                         ptrdiff_t stride)              \
-{                                                                       \
-    uint64_t half[8 + 9];                                               \
-    uint8_t * const halfH  = ((uint8_t*)half) + 64;                     \
-    uint8_t * const halfHV = ((uint8_t*)half);                          \
-    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
-                                                   stride, 9);          \
-    ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH, 8,           \
-                                        stride, 9);                     \
-    ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
-    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV,             \
-                                        stride, 8, 8);                  \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel8_mc31_ ## MMX(uint8_t *dst, uint8_t *src,    \
-                                         ptrdiff_t stride)              \
-{                                                                       \
-    uint64_t half[8 + 9];                                               \
-    uint8_t * const halfH  = ((uint8_t*)half) + 64;                     \
-    uint8_t * const halfHV = ((uint8_t*)half);                          \
-    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
-                                                   stride, 9);          \
-    ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src + 1, halfH, 8,       \
-                                        stride, 9);                     \
-    ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
-    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV,             \
-                                        stride, 8, 8);                  \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel8_mc13_ ## MMX(uint8_t *dst, uint8_t *src,    \
-                                         ptrdiff_t stride)              \
-{                                                                       \
-    uint64_t half[8 + 9];                                               \
-    uint8_t * const halfH  = ((uint8_t*)half) + 64;                     \
-    uint8_t * const halfHV = ((uint8_t*)half);                          \
-    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
-                                                   stride, 9);          \
-    ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH, 8,           \
-                                        stride, 9);                     \
-    ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
-    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH + 8, halfHV,         \
-                                        stride, 8, 8);                  \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel8_mc33_ ## MMX(uint8_t *dst, uint8_t *src,    \
-                                         ptrdiff_t stride)              \
-{                                                                       \
-    uint64_t half[8 + 9];                                               \
-    uint8_t * const halfH  = ((uint8_t*)half) + 64;                     \
-    uint8_t * const halfHV = ((uint8_t*)half);                          \
-    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
-                                                   stride, 9);          \
-    ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src + 1, halfH, 8,       \
-                                        stride, 9);                     \
-    ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
-    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH + 8, halfHV,         \
-                                        stride, 8, 8);                  \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel8_mc21_ ## MMX(uint8_t *dst, uint8_t *src,    \
-                                         ptrdiff_t stride)              \
-{                                                                       \
-    uint64_t half[8 + 9];                                               \
-    uint8_t * const halfH  = ((uint8_t*)half) + 64;                     \
-    uint8_t * const halfHV = ((uint8_t*)half);                          \
-    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
-                                                   stride, 9);          \
-    ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
-    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV,             \
-                                        stride, 8, 8);                  \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel8_mc23_ ## MMX(uint8_t *dst, uint8_t *src,    \
-                                         ptrdiff_t stride)              \
-{                                                                       \
-    uint64_t half[8 + 9];                                               \
-    uint8_t * const halfH  = ((uint8_t*)half) + 64;                     \
-    uint8_t * const halfHV = ((uint8_t*)half);                          \
-    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
-                                                   stride, 9);          \
-    ff_put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
-    ff_ ## OPNAME ## pixels8_l2_ ## MMX(dst, halfH + 8, halfHV,         \
-                                        stride, 8, 8);                  \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel8_mc12_ ## MMX(uint8_t *dst, uint8_t *src,    \
-                                         ptrdiff_t stride)              \
-{                                                                       \
-    uint64_t half[8 + 9];                                               \
-    uint8_t * const halfH = ((uint8_t*)half);                           \
-    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
-                                                   stride, 9);          \
-    ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH,              \
-                                        8, stride, 9);                  \
-    ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH,          \
-                                                   stride, 8);          \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel8_mc32_ ## MMX(uint8_t *dst, uint8_t *src,    \
-                                         ptrdiff_t stride)              \
-{                                                                       \
-    uint64_t half[8 + 9];                                               \
-    uint8_t * const halfH = ((uint8_t*)half);                           \
-    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
-                                                   stride, 9);          \
-    ff_put ## RND ## pixels8_l2_ ## MMX(halfH, src + 1, halfH, 8,       \
-                                        stride, 9);                     \
-    ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH,          \
-                                                   stride, 8);          \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel8_mc22_ ## MMX(uint8_t *dst, uint8_t *src,    \
-                                         ptrdiff_t stride)              \
-{                                                                       \
-    uint64_t half[9];                                                   \
-    uint8_t * const halfH = ((uint8_t*)half);                           \
-    ff_put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8,       \
-                                                   stride, 9);          \
-    ff_ ## OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH,          \
-                                                   stride, 8);          \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel16_mc00_ ## MMX (uint8_t *dst, uint8_t *src,  \
-                                           ptrdiff_t stride)            \
-{                                                                       \
-    ff_ ## OPNAME ## pixels16_ ## MMX(dst, src, stride, 16);            \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel16_mc10_ ## MMX(uint8_t *dst, uint8_t *src,   \
-                                          ptrdiff_t stride)             \
-{                                                                       \
-    uint64_t temp[32];                                                  \
-    uint8_t * const half = (uint8_t*)temp;                              \
-    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(half, src, 16,      \
-                                                    stride, 16);        \
-    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src, half, stride,        \
-                                         stride, 16);                   \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel16_mc20_ ## MMX(uint8_t *dst, uint8_t *src,   \
-                                          ptrdiff_t stride)             \
-{                                                                       \
-    ff_ ## OPNAME ## mpeg4_qpel16_h_lowpass_ ## MMX(dst, src,           \
-                                                    stride, stride, 16);\
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel16_mc30_ ## MMX(uint8_t *dst, uint8_t *src,   \
-                                          ptrdiff_t stride)             \
-{                                                                       \
-    uint64_t temp[32];                                                  \
-    uint8_t * const half = (uint8_t*)temp;                              \
-    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(half, src, 16,      \
-                                                    stride, 16);        \
-    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src + 1, half,            \
-                                         stride, stride, 16);           \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel16_mc01_ ## MMX(uint8_t *dst, uint8_t *src,   \
-                                          ptrdiff_t stride)             \
-{                                                                       \
-    uint64_t temp[32];                                                  \
-    uint8_t * const half = (uint8_t*)temp;                              \
-    ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(half, src, 16,      \
-                                                    stride);            \
-    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src, half, stride,        \
-                                         stride, 16);                   \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel16_mc02_ ## MMX(uint8_t *dst, uint8_t *src,   \
-                                          ptrdiff_t stride)             \
-{                                                                       \
-    ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, src,           \
-                                                    stride, stride);    \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel16_mc03_ ## MMX(uint8_t *dst, uint8_t *src,   \
-                                          ptrdiff_t stride)             \
-{                                                                       \
-    uint64_t temp[32];                                                  \
-    uint8_t * const half = (uint8_t*)temp;                              \
-    ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(half, src, 16,      \
-                                                    stride);            \
-    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, src+stride, half,         \
-                                         stride, stride, 16);           \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel16_mc11_ ## MMX(uint8_t *dst, uint8_t *src,   \
-                                          ptrdiff_t stride)             \
-{                                                                       \
-    uint64_t half[16 * 2 + 17 * 2];                                     \
-    uint8_t * const halfH  = ((uint8_t*)half) + 256;                    \
-    uint8_t * const halfHV = ((uint8_t*)half);                          \
-    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
-                                                    stride, 17);        \
-    ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16,         \
-                                         stride, 17);                   \
-    ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH,      \
-                                                    16, 16);            \
-    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV,            \
-                                         stride, 16, 16);               \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel16_mc31_ ## MMX(uint8_t *dst, uint8_t *src,   \
-                                          ptrdiff_t stride)             \
-{                                                                       \
-    uint64_t half[16 * 2 + 17 * 2];                                     \
-    uint8_t * const halfH  = ((uint8_t*)half) + 256;                    \
-    uint8_t * const halfHV = ((uint8_t*)half);                          \
-    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
-                                                    stride, 17);        \
-    ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src + 1, halfH, 16,     \
-                                         stride, 17);                   \
-    ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH,      \
-                                                    16, 16);            \
-    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV,            \
-                                         stride, 16, 16);               \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel16_mc13_ ## MMX(uint8_t *dst, uint8_t *src,   \
-                                          ptrdiff_t stride)             \
-{                                                                       \
-    uint64_t half[16 * 2 + 17 * 2];                                     \
-    uint8_t * const halfH  = ((uint8_t*)half) + 256;                    \
-    uint8_t * const halfHV = ((uint8_t*)half);                          \
-    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
-                                                    stride, 17);        \
-    ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16,         \
-                                         stride, 17);                   \
-    ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH,      \
-                                                    16, 16);            \
-    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH + 16, halfHV,       \
-                                         stride, 16, 16);               \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel16_mc33_ ## MMX(uint8_t *dst, uint8_t *src,   \
-                                          ptrdiff_t stride)             \
-{                                                                       \
-    uint64_t half[16 * 2 + 17 * 2];                                     \
-    uint8_t * const halfH  = ((uint8_t*)half) + 256;                    \
-    uint8_t * const halfHV = ((uint8_t*)half);                          \
-    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
-                                                    stride, 17);        \
-    ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src + 1, halfH, 16,     \
-                                         stride, 17);                   \
-    ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH,      \
-                                                    16, 16);            \
-    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH + 16, halfHV,       \
-                                         stride, 16, 16);               \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel16_mc21_ ## MMX(uint8_t *dst, uint8_t *src,   \
-                                          ptrdiff_t stride)             \
-{                                                                       \
-    uint64_t half[16 * 2 + 17 * 2];                                     \
-    uint8_t * const halfH  = ((uint8_t*)half) + 256;                    \
-    uint8_t * const halfHV = ((uint8_t*)half);                          \
-    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
-                                                    stride, 17);        \
-    ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH,      \
-                                                    16, 16);            \
-    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV,            \
-                                         stride, 16, 16);               \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel16_mc23_ ## MMX(uint8_t *dst, uint8_t *src,   \
-                                          ptrdiff_t stride)             \
-{                                                                       \
-    uint64_t half[16 * 2 + 17 * 2];                                     \
-    uint8_t * const halfH  = ((uint8_t*)half) + 256;                    \
-    uint8_t * const halfHV = ((uint8_t*)half);                          \
-    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
-                                                    stride, 17);        \
-    ff_put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH,      \
-                                                    16, 16);            \
-    ff_ ## OPNAME ## pixels16_l2_ ## MMX(dst, halfH + 16, halfHV,       \
-                                         stride, 16, 16);               \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel16_mc12_ ## MMX(uint8_t *dst, uint8_t *src,   \
-                                          ptrdiff_t stride)             \
-{                                                                       \
-    uint64_t half[17 * 2];                                              \
-    uint8_t * const halfH = ((uint8_t*)half);                           \
-    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
-                                                    stride, 17);        \
-    ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16,         \
-                                         stride, 17);                   \
-    ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH,         \
-                                                    stride, 16);        \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel16_mc32_ ## MMX(uint8_t *dst, uint8_t *src,   \
-                                          ptrdiff_t stride)             \
-{                                                                       \
-    uint64_t half[17 * 2];                                              \
-    uint8_t * const halfH = ((uint8_t*)half);                           \
-    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
-                                                    stride, 17);        \
-    ff_put ## RND ## pixels16_l2_ ## MMX(halfH, src + 1, halfH, 16,     \
-                                         stride, 17);                   \
-    ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH,         \
-                                                    stride, 16);        \
-}                                                                       \
-                                                                        \
-static void OPNAME ## qpel16_mc22_ ## MMX(uint8_t *dst, uint8_t *src,   \
-                                          ptrdiff_t stride)             \
-{                                                                       \
-    uint64_t half[17 * 2];                                              \
-    uint8_t * const halfH = ((uint8_t*)half);                           \
-    ff_put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16,     \
-                                                    stride, 17);        \
-    ff_ ## OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH,         \
-                                                    stride, 16);        \
-}
 
-QPEL_OP(put_,        _,        mmxext)
-QPEL_OP(avg_,        _,        mmxext)
-QPEL_OP(put_no_rnd_, _no_rnd_, mmxext)
-#endif /* HAVE_YASM */
-
-
-#if HAVE_INLINE_ASM
-static void gmc_mmx(uint8_t *dst, uint8_t *src,
-                    int stride, int h, int ox, int oy,
-                    int dxx, int dxy, int dyx, int dyy,
-                    int shift, int r, int width, int height)
+void ff_gmc_mmx(uint8_t *dst, uint8_t *src,
+                int stride, int h, int ox, int oy,
+                int dxx, int dxy, int dyx, int dyy,
+                int shift, int r, int width, int height)
 {
     const int w    = 8;
     const int ix   = ox  >> (16 + shift);
@@ -948,8 +499,8 @@ static void gmc_mmx(uint8_t *dst, uint8_t *src,
     }
 }
 
-static void vector_clipf_sse(float *dst, const float *src,
-                             float min, float max, int len)
+void ff_vector_clipf_sse(float *dst, const float *src,
+                         float min, float max, int len)
 {
     x86_reg i = (len - 16) * 4;
     __asm__ volatile (
@@ -983,262 +534,3 @@ static void vector_clipf_sse(float *dst, const float *src,
 }
 
 #endif /* HAVE_INLINE_ASM */
-
-void ff_h263_v_loop_filter_mmx(uint8_t *src, int stride, int qscale);
-void ff_h263_h_loop_filter_mmx(uint8_t *src, int stride, int qscale);
-
-int32_t ff_scalarproduct_int16_mmxext(const int16_t *v1, const int16_t *v2,
-                                      int order);
-int32_t ff_scalarproduct_int16_sse2(const int16_t *v1, const int16_t *v2,
-                                    int order);
-int32_t ff_scalarproduct_and_madd_int16_mmxext(int16_t *v1, const int16_t *v2,
-                                               const int16_t *v3,
-                                               int order, int mul);
-int32_t ff_scalarproduct_and_madd_int16_sse2(int16_t *v1, const int16_t *v2,
-                                             const int16_t *v3,
-                                             int order, int mul);
-int32_t ff_scalarproduct_and_madd_int16_ssse3(int16_t *v1, const int16_t *v2,
-                                              const int16_t *v3,
-                                              int order, int mul);
-
-void ff_apply_window_int16_round_mmxext(int16_t *output, const int16_t *input,
-                                        const int16_t *window, unsigned int len);
-void ff_apply_window_int16_round_sse2(int16_t *output, const int16_t *input,
-                                      const int16_t *window, unsigned int len);
-void ff_apply_window_int16_mmxext(int16_t *output, const int16_t *input,
-                                  const int16_t *window, unsigned int len);
-void ff_apply_window_int16_sse2(int16_t *output, const int16_t *input,
-                                const int16_t *window, unsigned int len);
-void ff_apply_window_int16_ssse3(int16_t *output, const int16_t *input,
-                                 const int16_t *window, unsigned int len);
-void ff_apply_window_int16_ssse3_atom(int16_t *output, const int16_t *input,
-                                      const int16_t *window, unsigned int len);
-
-void ff_bswap32_buf_ssse3(uint32_t *dst, const uint32_t *src, int w);
-void ff_bswap32_buf_sse2(uint32_t *dst, const uint32_t *src, int w);
-
-void ff_add_hfyu_median_prediction_mmxext(uint8_t *dst, const uint8_t *top,
-                                          const uint8_t *diff, int w,
-                                          int *left, int *left_top);
-int  ff_add_hfyu_left_prediction_ssse3(uint8_t *dst, const uint8_t *src,
-                                       int w, int left);
-int  ff_add_hfyu_left_prediction_sse4(uint8_t *dst, const uint8_t *src,
-                                      int w, int left);
-
-void ff_vector_clip_int32_mmx     (int32_t *dst, const int32_t *src,
-                                   int32_t min, int32_t max, unsigned int len);
-void ff_vector_clip_int32_sse2    (int32_t *dst, const int32_t *src,
-                                   int32_t min, int32_t max, unsigned int len);
-void ff_vector_clip_int32_int_sse2(int32_t *dst, const int32_t *src,
-                                   int32_t min, int32_t max, unsigned int len);
-void ff_vector_clip_int32_sse4    (int32_t *dst, const int32_t *src,
-                                   int32_t min, int32_t max, unsigned int len);
-
-#define SET_QPEL_FUNCS(PFX, IDX, SIZE, CPU, PREFIX)                          \
-    do {                                                                     \
-    c->PFX ## _pixels_tab[IDX][ 0] = PREFIX ## PFX ## SIZE ## _mc00_ ## CPU; \
-    c->PFX ## _pixels_tab[IDX][ 1] = PREFIX ## PFX ## SIZE ## _mc10_ ## CPU; \
-    c->PFX ## _pixels_tab[IDX][ 2] = PREFIX ## PFX ## SIZE ## _mc20_ ## CPU; \
-    c->PFX ## _pixels_tab[IDX][ 3] = PREFIX ## PFX ## SIZE ## _mc30_ ## CPU; \
-    c->PFX ## _pixels_tab[IDX][ 4] = PREFIX ## PFX ## SIZE ## _mc01_ ## CPU; \
-    c->PFX ## _pixels_tab[IDX][ 5] = PREFIX ## PFX ## SIZE ## _mc11_ ## CPU; \
-    c->PFX ## _pixels_tab[IDX][ 6] = PREFIX ## PFX ## SIZE ## _mc21_ ## CPU; \
-    c->PFX ## _pixels_tab[IDX][ 7] = PREFIX ## PFX ## SIZE ## _mc31_ ## CPU; \
-    c->PFX ## _pixels_tab[IDX][ 8] = PREFIX ## PFX ## SIZE ## _mc02_ ## CPU; \
-    c->PFX ## _pixels_tab[IDX][ 9] = PREFIX ## PFX ## SIZE ## _mc12_ ## CPU; \
-    c->PFX ## _pixels_tab[IDX][10] = PREFIX ## PFX ## SIZE ## _mc22_ ## CPU; \
-    c->PFX ## _pixels_tab[IDX][11] = PREFIX ## PFX ## SIZE ## _mc32_ ## CPU; \
-    c->PFX ## _pixels_tab[IDX][12] = PREFIX ## PFX ## SIZE ## _mc03_ ## CPU; \
-    c->PFX ## _pixels_tab[IDX][13] = PREFIX ## PFX ## SIZE ## _mc13_ ## CPU; \
-    c->PFX ## _pixels_tab[IDX][14] = PREFIX ## PFX ## SIZE ## _mc23_ ## CPU; \
-    c->PFX ## _pixels_tab[IDX][15] = PREFIX ## PFX ## SIZE ## _mc33_ ## CPU; \
-    } while (0)
-
-static av_cold void dsputil_init_mmx(DSPContext *c, AVCodecContext *avctx,
-                                     int mm_flags)
-{
-#if HAVE_MMX_INLINE
-    const int high_bit_depth = avctx->bits_per_raw_sample > 8;
-
-    c->put_pixels_clamped        = ff_put_pixels_clamped_mmx;
-    c->put_signed_pixels_clamped = ff_put_signed_pixels_clamped_mmx;
-    c->add_pixels_clamped        = ff_add_pixels_clamped_mmx;
-
-    if (!high_bit_depth) {
-        c->clear_block  = clear_block_mmx;
-        c->clear_blocks = clear_blocks_mmx;
-        c->draw_edges   = draw_edges_mmx;
-
-        switch (avctx->idct_algo) {
-        case FF_IDCT_AUTO:
-        case FF_IDCT_SIMPLEMMX:
-            c->idct_put              = ff_simple_idct_put_mmx;
-            c->idct_add              = ff_simple_idct_add_mmx;
-            c->idct                  = ff_simple_idct_mmx;
-            c->idct_permutation_type = FF_SIMPLE_IDCT_PERM;
-            break;
-        case FF_IDCT_XVIDMMX:
-            c->idct_put              = ff_idct_xvid_mmx_put;
-            c->idct_add              = ff_idct_xvid_mmx_add;
-            c->idct                  = ff_idct_xvid_mmx;
-            break;
-        }
-    }
-
-    c->gmc = gmc_mmx;
-
-    c->add_bytes = add_bytes_mmx;
-#endif /* HAVE_MMX_INLINE */
-
-#if HAVE_MMX_EXTERNAL
-    if (CONFIG_H263_DECODER || CONFIG_H263_ENCODER) {
-        c->h263_v_loop_filter = ff_h263_v_loop_filter_mmx;
-        c->h263_h_loop_filter = ff_h263_h_loop_filter_mmx;
-    }
-
-    c->vector_clip_int32 = ff_vector_clip_int32_mmx;
-#endif /* HAVE_MMX_EXTERNAL */
-}
-
-static av_cold void dsputil_init_mmxext(DSPContext *c, AVCodecContext *avctx,
-                                        int mm_flags)
-{
-#if HAVE_MMXEXT_INLINE
-    const int high_bit_depth = avctx->bits_per_raw_sample > 8;
-
-    if (!high_bit_depth && avctx->idct_algo == FF_IDCT_XVIDMMX) {
-        c->idct_put = ff_idct_xvid_mmxext_put;
-        c->idct_add = ff_idct_xvid_mmxext_add;
-        c->idct     = ff_idct_xvid_mmxext;
-    }
-#endif /* HAVE_MMXEXT_INLINE */
-
-#if HAVE_MMXEXT_EXTERNAL
-    SET_QPEL_FUNCS(avg_qpel,        0, 16, mmxext, );
-    SET_QPEL_FUNCS(avg_qpel,        1,  8, mmxext, );
-
-    SET_QPEL_FUNCS(put_qpel,        0, 16, mmxext, );
-    SET_QPEL_FUNCS(put_qpel,        1,  8, mmxext, );
-    SET_QPEL_FUNCS(put_no_rnd_qpel, 0, 16, mmxext, );
-    SET_QPEL_FUNCS(put_no_rnd_qpel, 1,  8, mmxext, );
-
-    /* slower than cmov version on AMD */
-    if (!(mm_flags & AV_CPU_FLAG_3DNOW))
-        c->add_hfyu_median_prediction = ff_add_hfyu_median_prediction_mmxext;
-
-    c->scalarproduct_int16          = ff_scalarproduct_int16_mmxext;
-    c->scalarproduct_and_madd_int16 = ff_scalarproduct_and_madd_int16_mmxext;
-
-    if (avctx->flags & CODEC_FLAG_BITEXACT) {
-        c->apply_window_int16 = ff_apply_window_int16_mmxext;
-    } else {
-        c->apply_window_int16 = ff_apply_window_int16_round_mmxext;
-    }
-#endif /* HAVE_MMXEXT_EXTERNAL */
-}
-
-static av_cold void dsputil_init_sse(DSPContext *c, AVCodecContext *avctx,
-                                     int mm_flags)
-{
-#if HAVE_SSE_INLINE
-    const int high_bit_depth = avctx->bits_per_raw_sample > 8;
-
-    if (!high_bit_depth) {
-        if (!(CONFIG_MPEG_XVMC_DECODER && avctx->xvmc_acceleration > 1)) {
-            /* XvMCCreateBlocks() may not allocate 16-byte aligned blocks */
-            c->clear_block  = clear_block_sse;
-            c->clear_blocks = clear_blocks_sse;
-        }
-    }
-
-    c->vector_clipf = vector_clipf_sse;
-#endif /* HAVE_SSE_INLINE */
-}
-
-static av_cold void dsputil_init_sse2(DSPContext *c, AVCodecContext *avctx,
-                                      int mm_flags)
-{
-#if HAVE_SSE2_INLINE
-    const int high_bit_depth = avctx->bits_per_raw_sample > 8;
-
-    if (!high_bit_depth && avctx->idct_algo == FF_IDCT_XVIDMMX) {
-        c->idct_put              = ff_idct_xvid_sse2_put;
-        c->idct_add              = ff_idct_xvid_sse2_add;
-        c->idct                  = ff_idct_xvid_sse2;
-        c->idct_permutation_type = FF_SSE2_IDCT_PERM;
-    }
-#endif /* HAVE_SSE2_INLINE */
-
-#if HAVE_SSE2_EXTERNAL
-    c->scalarproduct_int16          = ff_scalarproduct_int16_sse2;
-    c->scalarproduct_and_madd_int16 = ff_scalarproduct_and_madd_int16_sse2;
-    if (mm_flags & AV_CPU_FLAG_ATOM) {
-        c->vector_clip_int32 = ff_vector_clip_int32_int_sse2;
-    } else {
-        c->vector_clip_int32 = ff_vector_clip_int32_sse2;
-    }
-    if (avctx->flags & CODEC_FLAG_BITEXACT) {
-        c->apply_window_int16 = ff_apply_window_int16_sse2;
-    } else if (!(mm_flags & AV_CPU_FLAG_SSE2SLOW)) {
-        c->apply_window_int16 = ff_apply_window_int16_round_sse2;
-    }
-    c->bswap_buf = ff_bswap32_buf_sse2;
-#endif /* HAVE_SSE2_EXTERNAL */
-}
-
-static av_cold void dsputil_init_ssse3(DSPContext *c, AVCodecContext *avctx,
-                                       int mm_flags)
-{
-#if HAVE_SSSE3_EXTERNAL
-    c->add_hfyu_left_prediction = ff_add_hfyu_left_prediction_ssse3;
-    if (mm_flags & AV_CPU_FLAG_SSE4) // not really sse4, just slow on Conroe
-        c->add_hfyu_left_prediction = ff_add_hfyu_left_prediction_sse4;
-
-    if (mm_flags & AV_CPU_FLAG_ATOM)
-        c->apply_window_int16 = ff_apply_window_int16_ssse3_atom;
-    else
-        c->apply_window_int16 = ff_apply_window_int16_ssse3;
-    if (!(mm_flags & (AV_CPU_FLAG_SSE42|AV_CPU_FLAG_3DNOW))) // cachesplit
-        c->scalarproduct_and_madd_int16 = ff_scalarproduct_and_madd_int16_ssse3;
-    c->bswap_buf = ff_bswap32_buf_ssse3;
-#endif /* HAVE_SSSE3_EXTERNAL */
-}
-
-static av_cold void dsputil_init_sse4(DSPContext *c, AVCodecContext *avctx,
-                                      int mm_flags)
-{
-#if HAVE_SSE4_EXTERNAL
-    c->vector_clip_int32 = ff_vector_clip_int32_sse4;
-#endif /* HAVE_SSE4_EXTERNAL */
-}
-
-av_cold void ff_dsputil_init_mmx(DSPContext *c, AVCodecContext *avctx)
-{
-    int mm_flags = av_get_cpu_flags();
-
-#if HAVE_7REGS && HAVE_INLINE_ASM
-    if (mm_flags & AV_CPU_FLAG_CMOV)
-        c->add_hfyu_median_prediction = add_hfyu_median_prediction_cmov;
-#endif
-
-    if (mm_flags & AV_CPU_FLAG_MMX)
-        dsputil_init_mmx(c, avctx, mm_flags);
-
-    if (mm_flags & AV_CPU_FLAG_MMXEXT)
-        dsputil_init_mmxext(c, avctx, mm_flags);
-
-    if (mm_flags & AV_CPU_FLAG_SSE)
-        dsputil_init_sse(c, avctx, mm_flags);
-
-    if (mm_flags & AV_CPU_FLAG_SSE2)
-        dsputil_init_sse2(c, avctx, mm_flags);
-
-    if (mm_flags & AV_CPU_FLAG_SSSE3)
-        dsputil_init_ssse3(c, avctx, mm_flags);
-
-    if (mm_flags & AV_CPU_FLAG_SSE4)
-        dsputil_init_sse4(c, avctx, mm_flags);
-
-    if (CONFIG_ENCODERS)
-        ff_dsputilenc_init_mmx(c, avctx);
-}
index 388916b..c8615b2 100644 (file)
@@ -110,6 +110,27 @@ void ff_add_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels, int line_s
 void ff_put_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels, int line_size);
 void ff_put_signed_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels, int line_size);
 
+void ff_clear_block_mmx(int16_t *block);
+void ff_clear_block_sse(int16_t *block);
+void ff_clear_blocks_mmx(int16_t *blocks);
+void ff_clear_blocks_sse(int16_t *blocks);
+
+void ff_add_bytes_mmx(uint8_t *dst, uint8_t *src, int w);
+
+void ff_add_hfyu_median_prediction_cmov(uint8_t *dst, const uint8_t *top,
+                                        const uint8_t *diff, int w,
+                                        int *left, int *left_top);
+
+void ff_draw_edges_mmx(uint8_t *buf, int wrap, int width, int height,
+                       int w, int h, int sides);
+
+void ff_gmc_mmx(uint8_t *dst, uint8_t *src,
+                int stride, int h, int ox, int oy,
+                int dxx, int dxy, int dyx, int dyy,
+                int shift, int r, int width, int height);
+
+void ff_vector_clipf_sse(float *dst, const float *src,
+                         float min, float max, int len);
 
 void ff_avg_pixels8_mmx(uint8_t *block, const uint8_t *pixels,
                         ptrdiff_t line_size, int h);