fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-06 09:28:07 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

blorp: Emit STATE_COMPUTE_MODE before COMPUTE_WALKER

Browse source 
Cc: stable
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35563>
This commit is contained in:
José Roberto de Souza 2025-06-16 11:05:31 -07:00 • committed by Marge Bot
parent 52bd6fae0e
commit b37747ce68
1 changed files with 34 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

34
src/intel/blorp/blorp_genX_exec_brw.h
View file

@ -26,6 +26,7 @@
#include "blorp_priv.h"
#include "dev/intel_device_info.h"
#include "common/intel_common.h"
#include "common/intel_compute_slm.h"
#include "common/intel_sample_positions.h"
#include "common/intel_l3_config.h"
@ -1717,6 +1718,39 @@ blorp_exec_compute(struct blorp_batch *batch, const struct blorp_params *params)
assert(cs_prog_data->local_size[2] == 1);
#if GFX_VERx10 >= 125
/* Not need with VRT enabled */
#if GFX_VERx10 < 300
uint8_t pixel_async_compute_thread_limit, z_pass_async_compute_thread_limit,
np_z_async_throttle_settings;
bool slm_or_barrier_enabled = prog_data->total_shared != 0 || cs_prog_data->uses_barrier;
intel_compute_engine_async_threads_limit(devinfo, dispatch.threads,
slm_or_barrier_enabled,
&pixel_async_compute_thread_limit,
&z_pass_async_compute_thread_limit,
&np_z_async_throttle_settings);
blorp_emit(batch, GENX(STATE_COMPUTE_MODE), cm) {
#if GFX_VER >= 20
cm.AsyncComputeThreadLimit = pixel_async_compute_thread_limit;
cm.ZPassAsyncComputeThreadLimit = z_pass_async_compute_thread_limit;
cm.ZAsyncThrottlesettings = np_z_async_throttle_settings;
cm.AsyncComputeThreadLimitMask = 0x7;
cm.ZPassAsyncComputeThreadLimitMask = 0x7;
cm.ZAsyncThrottlesettingsMask = 0x3;
#else
cm.PixelAsyncComputeThreadLimit = pixel_async_compute_thread_limit;
cm.ZPassAsyncComputeThreadLimit = z_pass_async_compute_thread_limit;
cm.PixelAsyncComputeThreadLimitMask = 0x7;
cm.ZPassAsyncComputeThreadLimitMask = 0x7;
if (intel_device_info_is_mtl_or_arl(devinfo)) {
cm.ZAsyncThrottlesettings = np_z_async_throttle_settings;
cm.ZAsyncThrottlesettingsMask = 0x3;
}
#endif
}
#endif /* GFX_VERx10 < 300 */
uint32_t surfaces_offset = blorp_setup_binding_table(batch, params);
uint32_t samplers_offset =