svga: optimize constant buffer uploads

When a constant buffer slot is allocated in the upload buffer,
the allocated slot size is always in multiple of 256. But the actual buffer
size might not be in multiple of 256. This causes a gap between
the ending offset of a slot and the starting offset of the next slot.
The gap will prevent the two slots to be updated in a single update command.
In order to maximize the chance of merging the contiguous dirty ranges,
when a slot is to be allocated in the constant upload buffer,
specify a buffer size in multiple of 256.

There is about 10% performance improvement with Lightsmark2008 and
30% with Cinebench R11.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>

diff --git a/src/gallium/drivers/svga/svga_state_constants.c b/src/gallium/drivers/svga/svga_state_constants.c
index aee1ade..5ae0382 100644 (file)
--- a/src/gallium/drivers/svga/svga_state_constants.c
+++ b/src/gallium/drivers/svga/svga_state_constants.c
@@ -554,6 +554,7 @@ emit_constbuf_vgpu10(struct svga_context *svga, unsigned shader)
    void *src_map = NULL, *dst_map;
    unsigned offset;
    const struct svga_shader_variant *variant;
+   unsigned alloc_buf_size;
 
    assert(shader == PIPE_SHADER_VERTEX ||
           shader == PIPE_SHADER_GEOMETRY ||
@@ -618,7 +619,16 @@ emit_constbuf_vgpu10(struct svga_context *svga, unsigned shader)
     */
    new_buf_size = align(new_buf_size, 16);
 
-   u_upload_alloc(svga->const0_upload, 0, new_buf_size,
+   /* Constant buffer size in the upload buffer must be in multiples of 256.
+    * In order to maximize the chance of merging the upload buffer chunks
+    * when svga_buffer_add_range() is called,
+    * the allocate buffer size needs to be in multiples of 256 as well.
+    * Otherwise, since there is gap between each dirty range of the upload buffer,
+    * each dirty range will end up in its own UPDATE_GB_IMAGE command.
+    */
+   alloc_buf_size = align(new_buf_size, CONST0_UPLOAD_ALIGNMENT);
+
+   u_upload_alloc(svga->const0_upload, 0, alloc_buf_size,
                   CONST0_UPLOAD_ALIGNMENT, &offset,
                   &dst_buffer, &dst_map);
    if (!dst_map) {