x86: AVX2 predict_16x16_p

[x264] / common / common.h

diff --git a/common/common.h b/common/common.h
index 39ad5cb1527d7d674aa16efce7e2dd14bc2736da..53a6ff03c0b834f592867fbcb348e3a8e8a89649 100644 (file)
--- a/common/common.h
+++ b/common/common.h
@@ -40,6 +40,7 @@
 #define IS_DISPOSABLE(type) ( type == X264_TYPE_B )
 #define FIX8(f) ((int)(f*(1<<8)+.5))
 #define ALIGN(x,a) (((x)+((a)-1))&~((a)-1))
+#define ARRAY_ELEMS(a) ((sizeof(a))/(sizeof(a[0])))
 
 #define CHECKED_MALLOC( var, size )\
 do {\
@@ -53,6 +54,8 @@ do {\
     memset( var, 0, size );\
 } while( 0 )
 
+#define ARRAY_SIZE(array)  (sizeof(array)/sizeof(array[0]))
+
 #define X264_BFRAME_MAX 16
 #define X264_REF_MAX 16
 #define X264_THREAD_MAX 128
@@ -93,6 +96,10 @@ do {\
 #include <assert.h>
 #include <limits.h>
 
+#if HAVE_OPENCL
+#include "opencl.h"
+#endif
+
 #if HAVE_INTERLACED
 #   define MB_INTERLACED h->mb.b_interlaced
 #   define SLICE_MBAFF h->sh.b_mbaff
@@ -202,6 +209,7 @@ static const uint8_t x264_scan8[16*3 + 3] =
 };
 
 #include "x264.h"
+#include "cabac.h"
 #include "bitstream.h"
 #include "set.h"
 #include "predict.h"
@@ -209,7 +217,6 @@ static const uint8_t x264_scan8[16*3 + 3] =
 #include "mc.h"
 #include "frame.h"
 #include "dct.h"
-#include "cabac.h"
 #include "quant.h"
 #include "cpu.h"
 #include "threadpool.h"
@@ -603,11 +610,11 @@ struct x264_t
     /* Current MB DCT coeffs */
     struct
     {
-        ALIGNED_16( dctcoef luma16x16_dc[3][16] );
+        ALIGNED_N( dctcoef luma16x16_dc[3][16] );
         ALIGNED_16( dctcoef chroma_dc[2][8] );
         // FIXME share memory?
-        ALIGNED_16( dctcoef luma8x8[12][64] );
-        ALIGNED_16( dctcoef luma4x4[16*3][16] );
+        ALIGNED_N( dctcoef luma8x8[12][64] );
+        ALIGNED_N( dctcoef luma4x4[16*3][16] );
     } dct;
 
     /* MB table and cache for current frame/mb */
@@ -746,7 +753,7 @@ struct x264_t
 #define FENC_STRIDE 16
 #define FDEC_STRIDE 32
             ALIGNED_16( pixel fenc_buf[48*FENC_STRIDE] );
-            ALIGNED_16( pixel fdec_buf[52*FDEC_STRIDE] );
+            ALIGNED_N( pixel fdec_buf[52*FDEC_STRIDE] );
 
             /* i4x4 and i8x8 backup data, for skipping the encode stage when possible */
             ALIGNED_16( pixel i4x4_fdec_buf[16*16] );
@@ -935,6 +942,10 @@ struct x264_t
     struct visualize_t *visualize;
 #endif
     x264_lookahead_t *lookahead;
+
+#if HAVE_OPENCL
+    x264_opencl_t opencl;
+#endif
 };
 
 // included at the end because it needs x264_t