ARMv6 optimized half pixel variance calculations

Adds following ARMv6 optimized functions to the encoder:
 - vp8_variance_halfpixvar16x16_h_armv6
 - vp8_variance_halfpixvar16x16_v_armv6
 - vp8_variance_halfpixvar16x16_hv_armv6

Change-Id: I1e9c2af7acd2a51b72b3845beecd990db4bebd29

diff --git a/vp8/encoder/arm/arm_csystemdependent.c b/vp8/encoder/arm/arm_csystemdependent.c
index 5852afddb7edc2069a7bfd7662bcbcd40fae6e2b..73007d4145d9bd01969448f901212bd243441e98 100644 (file)
--- a/vp8/encoder/arm/arm_csystemdependent.c
+++ b/vp8/encoder/arm/arm_csystemdependent.c
@@ -46,6 +46,9 @@ void vp8_arch_arm_encoder_init(VP8_COMP *cpi)
         cpi->rtcd.variance.subpixvar8x16         = vp8_sub_pixel_variance8x16_c;
         cpi->rtcd.variance.subpixvar16x8         = vp8_sub_pixel_variance16x8_c;*/
         cpi->rtcd.variance.subpixvar16x16        = vp8_sub_pixel_variance16x16_armv6;
+        cpi->rtcd.variance.halfpixvar16x16_h     = vp8_variance_halfpixvar16x16_h_armv6;
+        cpi->rtcd.variance.halfpixvar16x16_v     = vp8_variance_halfpixvar16x16_v_armv6;
+        cpi->rtcd.variance.halfpixvar16x16_hv    = vp8_variance_halfpixvar16x16_hv_armv6;
 
         /*cpi->rtcd.variance.mse16x16              = vp8_mse16x16_c;
         cpi->rtcd.variance.getmbss               = vp8_get_mb_ss_c;*/

diff --git a/vp8/encoder/arm/variance_arm.c b/vp8/encoder/arm/variance_arm.c
index 1cf9fd8f00d5d55e6b751d54e056907ef95dbe40..64d76bcf8685e5791357bad9c0078ef56a3d29c0 100644 (file)
--- a/vp8/encoder/arm/variance_arm.c
+++ b/vp8/encoder/arm/variance_arm.c
@@ -43,7 +43,41 @@ unsigned int vp8_sub_pixel_variance16x16_armv6
                                    dst_pixels_per_line, sse);
 }
 
-#endif
+unsigned int vp8_variance_halfpixvar16x16_h_armv6(
+    const unsigned char *src_ptr,
+    int  source_stride,
+    const unsigned char *ref_ptr,
+    int  recon_stride,
+    unsigned int *sse)
+{
+    return vp8_sub_pixel_variance16x16_armv6(src_ptr, source_stride, 4, 0,
+                                         ref_ptr, recon_stride, sse);
+}
+
+unsigned int vp8_variance_halfpixvar16x16_v_armv6(
+    const unsigned char *src_ptr,
+    int  source_stride,
+    const unsigned char *ref_ptr,
+    int  recon_stride,
+    unsigned int *sse)
+{
+    return vp8_sub_pixel_variance16x16_armv6(src_ptr, source_stride, 0, 4,
+                                         ref_ptr, recon_stride, sse);
+}
+
+unsigned int vp8_variance_halfpixvar16x16_hv_armv6(
+    const unsigned char *src_ptr,
+    int  source_stride,
+    const unsigned char *ref_ptr,
+    int  recon_stride,
+    unsigned int *sse)
+{
+    return vp8_sub_pixel_variance16x16_armv6(src_ptr, source_stride, 4, 4,
+                                         ref_ptr, recon_stride, sse);
+}
+
+#endif /* HAVE_ARMV6 */
+
 
 #if HAVE_ARMV7
 

diff --git a/vp8/encoder/arm/variance_arm.h b/vp8/encoder/arm/variance_arm.h
index c807e29c06750d843c9408bcb589b2a41d471388..7ac0ac08e3fb7270560b65109c779306f6047472 100644 (file)
--- a/vp8/encoder/arm/variance_arm.h
+++ b/vp8/encoder/arm/variance_arm.h
@@ -17,6 +17,9 @@
 extern prototype_sad(vp8_sad16x16_armv6);
 extern prototype_variance(vp8_variance16x16_armv6);
 extern prototype_subpixvariance(vp8_sub_pixel_variance16x16_armv6);
+extern prototype_variance(vp8_variance_halfpixvar16x16_h_armv6);
+extern prototype_variance(vp8_variance_halfpixvar16x16_v_armv6);
+extern prototype_variance(vp8_variance_halfpixvar16x16_hv_armv6);
 
 #if !CONFIG_RUNTIME_CPU_DETECT
 
@@ -29,10 +32,20 @@ extern prototype_subpixvariance(vp8_sub_pixel_variance16x16_armv6);
 #undef  vp8_variance_var16x16
 #define vp8_variance_var16x16 vp8_variance16x16_armv6
 
+#undef  vp8_variance_halfpixvar16x16_h
+#define vp8_variance_halfpixvar16x16_h vp8_variance_halfpixvar16x16_h_armv6
+
+#undef  vp8_variance_halfpixvar16x16_v
+#define vp8_variance_halfpixvar16x16_v vp8_variance_halfpixvar16x16_v_armv6
+
+#undef  vp8_variance_halfpixvar16x16_hv
+#define vp8_variance_halfpixvar16x16_hv vp8_variance_halfpixvar16x16_hv_armv6
+
 #endif /* !CONFIG_RUNTIME_CPU_DETECT */
 
 #endif /* HAVE_ARMV6 */
 
+
 #if HAVE_ARMV7
 extern prototype_sad(vp8_sad4x4_neon);
 extern prototype_sad(vp8_sad8x8_neon);