fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-04 20:38:06 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 2

radv: rework the optimal packet order for dispatches

Browse source 
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37013>
This commit is contained in:
Samuel Pitoiset 2025-08-29 13:53:43 +02:00 • committed by Marge Bot
parent 62c92a0516
commit 7f12f98741
1 changed files with 10 additions and 35 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

45
src/amd/vulkan/radv_cmd_buffer.c
View file

@ -13368,49 +13368,24 @@ radv_before_dispatch(struct radv_cmd_buffer *cmd_buffer, struct radv_compute_pip
struct radv_shader *compute_shader = bind_point == VK_PIPELINE_BIND_POINT_COMPUTE
? cmd_buffer->state.shaders[MESA_SHADER_COMPUTE]
: cmd_buffer->state.rt_prolog;
const bool has_prefetch = pdev->info.gfx_level >= GFX7;
if (compute_shader->info.cs.regalloc_hang_bug)
cmd_buffer->state.flush_bits |= RADV_CMD_FLAG_PS_PARTIAL_FLUSH | RADV_CMD_FLAG_CS_PARTIAL_FLUSH;
if (cmd_buffer->state.flush_bits & (RADV_CMD_FLAG_FLUSH_AND_INV_CB | RADV_CMD_FLAG_FLUSH_AND_INV_DB |
RADV_CMD_FLAG_PS_PARTIAL_FLUSH | RADV_CMD_FLAG_CS_PARTIAL_FLUSH)) {
/* If we have to wait for idle, set all states first, so that
* all SET packets are processed in parallel with previous draw
* calls. Then upload descriptors, set shader pointers, and
* dispatch, and prefetch at the end. This ensures that the
* time the CUs are idle is very short. (there are only SET_SH
* packets between the wait and the draw)
*/
if (pipeline)
radv_emit_compute_pipeline(cmd_buffer, pipeline);
if (bind_point == VK_PIPELINE_BIND_POINT_RAY_TRACING_KHR)
radv_emit_rt_stack_size(cmd_buffer);
radv_emit_cache_flush(cmd_buffer);
/* <-- CUs are idle here --> */
/* Use the optimal packet order similar to draws. */
if (pipeline)
radv_emit_compute_pipeline(cmd_buffer, pipeline);
if (bind_point == VK_PIPELINE_BIND_POINT_RAY_TRACING_KHR)
radv_emit_rt_stack_size(cmd_buffer);
radv_upload_compute_shader_descriptors(cmd_buffer, bind_point);
} else {
/* If we don't wait for idle, start prefetches first, then set
* states, and dispatch at the end.
*/
radv_emit_cache_flush(cmd_buffer);
radv_upload_compute_shader_descriptors(cmd_buffer, bind_point);
if (has_prefetch) {
if (bind_point == VK_PIPELINE_BIND_POINT_COMPUTE) {
radv_emit_compute_prefetch(cmd_buffer);
} else {
radv_emit_ray_tracing_prefetch(cmd_buffer);
}
}
if (pdev->info.gfx_level >= GFX12)
radv_gfx12_emit_buffered_regs(device, cmd_buffer->cs);
radv_upload_compute_shader_descriptors(cmd_buffer, bind_point);
radv_emit_cache_flush(cmd_buffer);
if (pipeline)
radv_emit_compute_pipeline(cmd_buffer, pipeline);
if (bind_point == VK_PIPELINE_BIND_POINT_RAY_TRACING_KHR)
radv_emit_rt_stack_size(cmd_buffer);
}
/* <-- CUs are idle here if shaders are synchronized. */
if (pipeline_is_dirty) {
/* Raytracing uses compute shaders but has separate bind points and pipelines.