fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-01-07 21:40:20 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

nvk: Use MEM_LOCAL for nvk_cmd_mem_create

Browse source 
It is safe to allocate command buffer memory as MEM_LOCAL (VRAM or GART)
when we can map GPU memory (we trust NVKMD to force GART).

This reduces latency on pyrowave for compute dispatch (~1.90ms -> ~1.48ms),
likely as QMD and the root tables should not have been in GART in the first place.

Signed-off-by: Mary Guillemard <mary@mary.zone>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36970>
This commit is contained in:
Mary Guillemard 2025-08-23 20:11:01 +02:00 • committed by Marge Bot
parent 5d77057ff6
commit becccebd21
1 changed files with 3 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

6
src/nouveau/vulkan/nvk_cmd_pool.c
View file

@ -21,9 +21,9 @@ nvk_cmd_mem_create(struct nvk_cmd_pool *pool, bool force_gart, struct nvk_cmd_me
if (mem == NULL)
return vk_error(pool, VK_ERROR_OUT_OF_HOST_MEMORY);
uint32_t flags = NVKMD_MEM_GART;
if (force_gart)
assert(flags & NVKMD_MEM_GART);
const uint32_t flags = force_gart ? NVKMD_MEM_GART
: NVKMD_MEM_LOCAL;
result = nvkmd_dev_alloc_mapped_mem(dev->nvkmd, &pool->vk.base,
NVK_CMD_MEM_SIZE, 0,
flags, NVKMD_MEM_MAP_WR,