fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-01-08 10:40:11 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

winsys/amdgpu: bypass GL2 for command buffers

Browse source 
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26307>
This commit is contained in:
Marek Olšák 2023-11-25 07:41:24 -05:00 • committed by Marge Bot
parent c77bcf00a3
commit e1e35112c8
1 changed files with 5 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

6
src/gallium/winsys/amdgpu/drm/amdgpu_cs.c
View file

@ -871,9 +871,13 @@ static bool amdgpu_ib_new_buffer(struct amdgpu_winsys *ws,
/* Use cached GTT for command buffers. Writing to other heaps is very slow on the CPU.
* The speed of writing to GTT WC is somewhere between no difference and very slow, while
* VRAM being very slow a lot more often.
*
* Bypass GL2 because command buffers are read only once. Bypassing GL2 has better latency
* and doesn't have to wait for cached GL2 requests to be processed.
*/
enum radeon_bo_domain domain = RADEON_DOMAIN_GTT;
unsigned flags = RADEON_FLAG_NO_INTERPROCESS_SHARING;
unsigned flags = RADEON_FLAG_NO_INTERPROCESS_SHARING |
RADEON_FLAG_GL2_BYPASS;
if (cs->ip_type == AMD_IP_GFX ||
cs->ip_type == AMD_IP_COMPUTE ||