shithub: libvpx

Download patch

ref: afee58f2c4159172f5340f2c7d3e8041cfa0eb91
parent: c39cd9235ece2e55f0cc84fb0f9f5e698146ee9c
author: Linfeng Zhang <linfengz@google.com>
date: Tue Sep 12 07:37:04 EDT 2017 

Specialize 4 to 3 scaling in vp9_scale_and_extend_frame_c()

Scale 3x3 block instead of 16x16 block in each loop.

Benefits:
1. Reduced number of different phase_scaler from 16 to 3. Optimization code
   will be smaller and faster.
2. The maximum phase_scaler drifting will be reduced from 5/16 to 1/24.
   (The drifting is 1/(3*16) in each step.)

BUG=webm:1419

Change-Id: Ibb9242a629ddb03e1ff93b859bece738255e698c


--- a/vp9/encoder/vp9_frame_scale.c
+++ b/vp9/encoder/vp9_frame_scale.c
@@ -20,8 +20,6 @@
                                   INTERP_FILTER filter_type, int phase_scaler) {
   const int src_w = src->y_crop_width;
   const int src_h = src->y_crop_height;
-  const int dst_w = dst->y_crop_width;
-  const int dst_h = dst->y_crop_height;
   const uint8_t *const srcs[3] = { src->y_buffer, src->u_buffer,
                                    src->v_buffer };
   const int src_strides[3] = { src->y_stride, src->uv_stride, src->uv_stride };
@@ -30,22 +28,48 @@
   const InterpKernel *const kernel = vp9_filter_kernels[filter_type];
   int x, y, i;
 
-  for (i = 0; i < MAX_MB_PLANE; ++i) {
-    const int factor = (i == 0 || i == 3 ? 1 : 2);
-    const int src_stride = src_strides[i];
-    const int dst_stride = dst_strides[i];
-    for (y = 0; y < dst_h; y += 16) {
-      const int y_q4 = y * (16 / factor) * src_h / dst_h + phase_scaler;
-      for (x = 0; x < dst_w; x += 16) {
-        const int x_q4 = x * (16 / factor) * src_w / dst_w + phase_scaler;
-        const uint8_t *src_ptr = srcs[i] +
-                                 (y / factor) * src_h / dst_h * src_stride +
-                                 (x / factor) * src_w / dst_w;
-        uint8_t *dst_ptr = dsts[i] + (y / factor) * dst_stride + (x / factor);
+  if (4 * dst->y_crop_width == 3 * src_w &&
+      4 * dst->y_crop_height == 3 * src_h) {
+    // Specialize 4 to 3 scaling.
+    const int dst_ws[3] = { dst->y_crop_width, dst->uv_crop_width,
+                            dst->uv_crop_width };
+    const int dst_hs[3] = { dst->y_crop_height, dst->uv_crop_height,
+                            dst->uv_crop_height };
+    for (i = 0; i < MAX_MB_PLANE; ++i) {
+      const int dst_w = dst_ws[i];
+      const int dst_h = dst_hs[i];
+      const int src_stride = src_strides[i];
+      const int dst_stride = dst_strides[i];
+      for (y = 0; y < dst_h; y += 3) {
+        for (x = 0; x < dst_w; x += 3) {
+          const uint8_t *src_ptr = srcs[i] + 4 * y / 3 * src_stride + 4 * x / 3;
+          uint8_t *dst_ptr = dsts[i] + y * dst_stride + x;
 
-        vpx_scaled_2d(src_ptr, src_stride, dst_ptr, dst_stride, kernel,
-                      x_q4 & 0xf, 16 * src_w / dst_w, y_q4 & 0xf,
-                      16 * src_h / dst_h, 16 / factor, 16 / factor);
+          vpx_scaled_2d_c(src_ptr, src_stride, dst_ptr, dst_stride, kernel,
+                          phase_scaler, 64 / 3, phase_scaler, 64 / 3, 3, 3);
+        }
+      }
+    }
+  } else {
+    const int dst_w = dst->y_crop_width;
+    const int dst_h = dst->y_crop_height;
+    for (i = 0; i < MAX_MB_PLANE; ++i) {
+      const int factor = (i == 0 || i == 3 ? 1 : 2);
+      const int src_stride = src_strides[i];
+      const int dst_stride = dst_strides[i];
+      for (y = 0; y < dst_h; y += 16) {
+        const int y_q4 = y * (16 / factor) * src_h / dst_h + phase_scaler;
+        for (x = 0; x < dst_w; x += 16) {
+          const int x_q4 = x * (16 / factor) * src_w / dst_w + phase_scaler;
+          const uint8_t *src_ptr = srcs[i] +
+                                   (y / factor) * src_h / dst_h * src_stride +
+                                   (x / factor) * src_w / dst_w;
+          uint8_t *dst_ptr = dsts[i] + (y / factor) * dst_stride + (x / factor);
+
+          vpx_scaled_2d(src_ptr, src_stride, dst_ptr, dst_stride, kernel,
+                        x_q4 & 0xf, 16 * src_w / dst_w, y_q4 & 0xf,
+                        16 * src_h / dst_h, 16 / factor, 16 / factor);
+        }
       }
     }
   }
-- 
⑨