Mesa (master): gallium: Use unpack_rgba() instead of fetch_rgba in translate_generic

[GitLab Mirror]

Module: Mesa
Branch: master
Commit: 09196355b2b2b6a2d3935eb3c43fe784d226426a
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=09196355b2b2b6a2d3935eb3c43fe784d226426a

Author: Eric Anholt <eric at anholt.net>
Date:   Thu Aug 13 10:14:11 2020 -0700

gallium: Use unpack_rgba() instead of fetch_rgba in translate_generic

This is the only user of fetch_rgba outside of llvmpipe, and it's in the
fallback path of this fallback path.  Looking at an example of these two
functions, b8g8r8a8's unpack_rgba is 2.7x as long as fetch_rgba.  It feels
reasonable to sacrifice some perf in this already slow (VBO readback, and
a function pointer call per attribute per vertex) path to reduce our
binary size.  And, if I ever finish getting unpack codegen to switch to
rows instead of rects, that factor will go back down.

Saves 40kb of binary on non-llvmpipe gallium drivers.

Reviewed-by: Marek Olšák <marek.olsak at amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6305>

---

 src/gallium/auxiliary/translate/translate_generic.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/src/gallium/auxiliary/translate/translate_generic.c b/src/gallium/auxiliary/translate/translate_generic.c
index 14631da33cd..d48c8f96655 100644
--- a/src/gallium/auxiliary/translate/translate_generic.c
+++ b/src/gallium/auxiliary/translate/translate_generic.c
@@ -50,7 +50,9 @@ struct translate_generic {
    struct {
       enum translate_element_type type;
 
-      util_format_fetch_rgba_func_ptr fetch;
+      void (*fetch)(void *dst, unsigned dst_stride,
+                    const uint8_t *src, unsigned src_stride,
+                    unsigned width, unsigned height);
       unsigned buffer;
       unsigned input_offset;
       unsigned instance_divisor;
@@ -623,7 +625,7 @@ generic_run_one(struct translate_generic *tg,
          if (likely(copy_size >= 0)) {
             memcpy(dst, src, copy_size);
          } else {
-            tg->attrib[attr].fetch(data, src, 0, 0);
+            tg->attrib[attr].fetch(data, 0, src, 0, 1, 1);
 
             if (0)
                debug_printf("Fetch linear attr %d  from %p  stride %d  index %d: "
@@ -796,6 +798,8 @@ translate_generic_create(const struct translate_key *key)
    for (i = 0; i < key->nr_elements; i++) {
       const struct util_format_description *format_desc =
             util_format_description(key->element[i].input_format);
+      const struct util_format_unpack_description *unpack =
+         util_format_unpack_description(key->element[i].input_format);
 
       assert(format_desc);
 
@@ -811,8 +815,7 @@ translate_generic_create(const struct translate_key *key)
          }
       }
 
-      tg->attrib[i].fetch =
-         util_format_fetch_rgba_func(key->element[i].input_format);
+      tg->attrib[i].fetch = unpack->unpack_rgba;
       tg->attrib[i].buffer = key->element[i].input_buffer;
       tg->attrib[i].input_offset = key->element[i].input_offset;
       tg->attrib[i].instance_divisor = key->element[i].instance_divisor;