fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-06 02:58:05 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

pan/bi: Rework mem_vectorize_cb

Browse source 
Intstead of focusing on numbers of components and bit sizes, focus on
the total number of bytes read.

Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Reviewed-by: Christoph Pillmayer <christoph.pillmayer@arm.com>
Reviewed-by: Lorenzo Rossi <lorenzo.rossi@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40576>
This commit is contained in:
Faith Ekstrand 2026-03-23 15:23:15 -04:00 • committed by Marge Bot
parent a5801b1a23
commit c2fc7d49e8
1 changed files with 5 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

13
src/panfrost/compiler/bifrost/bifrost_compile.c
View file

@ -5576,18 +5576,15 @@ mem_vectorize_cb(unsigned align_mul, unsigned align_offset, unsigned bit_size,
if (hole_size > 0)
return false;
/* Must be aligned to the size of the load */
unsigned align = nir_combined_align(align_mul, align_offset);
if ((bit_size / 8) > align)
return false;
/* We have a hard limit of at most 4 components */
if (num_components > 4)
return false;
if (bit_size > 32)
return false;
const unsigned bytes = num_components * (bit_size / 8);
const unsigned max_bytes = 128u / 8u; /* LOAD.i128 */
return true;
const unsigned combined_align = nir_combined_align(align_mul, align_offset);
return bytes <= combined_align && bytes <= max_bytes;
}
static void