Patch to make WashingtonDC *not* copy framebuffers upside-down. It actually slows things down because it causes more cache misses.

upside_down.diff

diff --git a/src/gfx/opengl/opengl_output.c b/src/gfx/opengl/opengl_output.c
index e00a4df..172596f 100644
--- a/src/gfx/opengl/opengl_output.c
+++ b/src/gfx/opengl/opengl_output.c
@@ -56,17 +56,11 @@ static struct shader fb_shader;
 #define FB_VERT_LEN 5
 #define FB_VERT_COUNT 4
 static GLfloat fb_quad_verts[FB_VERT_LEN * FB_VERT_COUNT] = {
-    /*
-     * it is not a mistake that the texture-coordinates are upside-down
-     * this is because dreamcast puts the origin at upper-left corner,
-     * but opengl textures put the origin at the lower-left corner
-     */
-
     // position            // texture coordinates
-    -1.0f, -1.0f, 0.0f,    0.0f, 1.0f,
-    -1.0f,  1.0f, 0.0f,    0.0f, 0.0f,
-     1.0f,  1.0f, 0.0f,    1.0f, 0.0f,
-     1.0f, -1.0f, 0.0f,    1.0f, 1.0f
+    -1.0f,  1.0f, 0.0f,    0.0f, 1.0f,
+    -1.0f, -1.0f, 0.0f,    0.0f, 0.0f,
+     1.0f, -1.0f, 0.0f,    1.0f, 0.0f,
+     1.0f,  1.0f, 0.0f,    1.0f, 1.0f
 };
 
 #define FB_QUAD_IDX_COUNT 4
diff --git a/src/hw/pvr2/framebuffer.c b/src/hw/pvr2/framebuffer.c
index 427a6a1..758b2d0 100644
--- a/src/hw/pvr2/framebuffer.c
+++ b/src/hw/pvr2/framebuffer.c
@@ -184,7 +184,7 @@ void read_framebuffer_rgb565_prog(uint32_t *pixels_out, addr32_t start_addr,
     unsigned row;
     for (row = 0; row < height; row++) {
         uint16_t const *in_col_start = pixels_in + stride * row;
-        uint32_t *out_col_start = pixels_out + row * width;
+        uint32_t *out_col_start = pixels_out + (height - row - 1) * width;
 
         conv_rgb565_to_rgba8888(out_col_start, in_col_start, width, concat);
     }
@@ -246,17 +246,16 @@ void read_framebuffer_rgb565_intl(uint32_t *pixels_out,
     unsigned row;
     for (row = 0; row < fb_height; row++) {
         uint16_t const *ptr_row1 =
-            (uint16_t const*)(pvr2_tex32_mem + row_start_field1);
+            (uint16_t const*)(pvr2_tex32_mem + row_start_field1 + field_adv * row);
         uint16_t const *ptr_row2 =
-            (uint16_t const*)(pvr2_tex32_mem + row_start_field2);
-
-        conv_rgb565_to_rgba8888(pixels_out + (row << 1) * fb_width,
-                                ptr_row1, fb_width, concat);
-        conv_rgb565_to_rgba8888(pixels_out + ((row << 1) + 1) * fb_width,
-                                ptr_row2, fb_width, concat);
-
-        row_start_field1 += field_adv;
-        row_start_field2 += field_adv;
+            (uint16_t const*)(pvr2_tex32_mem + row_start_field2 + field_adv * row);
+
+        conv_rgb565_to_rgba8888(
+            pixels_out + (((fb_height - row - 1) << 1) + 1) * fb_width,
+            ptr_row1, fb_width, concat);
+        conv_rgb565_to_rgba8888(
+            pixels_out + ((fb_height - row - 1) << 1) * fb_width,
+            ptr_row2, fb_width, concat);
     }
 }
 
@@ -287,7 +286,7 @@ void read_framebuffer_rgb0888_prog(uint32_t *pixels_out, addr32_t start_addr,
     unsigned row;
     for (row = 0; row < height; row++) {
         uint32_t const *in_col_start = pixels_in + width * row;
-        uint32_t *out_col_start = pixels_out + row * width;
+        uint32_t *out_col_start = pixels_out + (height - row - 1) * width;
 
         conv_rgb0888_to_rgba8888(out_col_start, in_col_start, width);
     }
@@ -338,17 +337,14 @@ void read_framebuffer_rgb0888_intl(uint32_t *pixels_out,
     unsigned row;
     for (row = 0; row < fb_height; row++) {
         uint32_t const *ptr_row1 =
-            (uint32_t const*)(pvr2_tex32_mem + row_start_field1);
+            (uint32_t const*)(pvr2_tex32_mem + row_start_field1 + field_adv * row);
         uint32_t const *ptr_row2 =
-            (uint32_t const*)(pvr2_tex32_mem + row_start_field2);
+            (uint32_t const*)(pvr2_tex32_mem + row_start_field2 + field_adv * row);
 
-        conv_rgb0888_to_rgba8888(pixels_out + (row << 1) * fb_width,
+        conv_rgb0888_to_rgba8888(pixels_out + (((fb_height - row - 1) << 1) + 1) * fb_width,
                                  ptr_row1, fb_width);
-        conv_rgb0888_to_rgba8888(pixels_out + ((row << 1) + 1) * fb_width,
+        conv_rgb0888_to_rgba8888(pixels_out + ((fb_height - row - 1) << 1) * fb_width,
                                  ptr_row2, fb_width);
-
-        row_start_field1 += field_adv;
-        row_start_field2 += field_adv;
     }
 }