fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-01-03 07:10:15 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

anv: Wa_18040903259 only applies to RCS when in GPGPU mode
Some checks are pending
macOS-CI / macOS-CI (dri) (push) Waiting to run
Details
macOS-CI / macOS-CI (xlib) (push) Waiting to run
Details

Browse source 
Sadly this probably won't change anything in terms of perf as the CCS
engine has a bunch of other restrictions.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 243c01c703 ("anv/iris: implement Wa_18040903259")
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38484>
This commit is contained in:
Lionel Landwerlin 2025-11-14 12:04:30 +02:00 • committed by Marge Bot
parent 168a8d0b52
commit 07b7de35cc
1 changed files with 18 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

25
src/intel/vulkan/genX_cmd_buffer.c
View file

@ -2445,16 +2445,27 @@ emit_pipe_control(struct anv_batch *batch,
trace_intel_begin_stall(batch->trace);
}
/* XXX - insert all workarounds and GFX specific things below. */
#if INTEL_WA_1607156449_GFX_VER || INTEL_NEEDS_WA_18040903259
#if INTEL_WA_1607156449_GFX_VER
/* Wa_1607156449: For COMPUTE Workload - Any PIPE_CONTROL command with
* POST_SYNC Operation Enabled MUST be preceded by a PIPE_CONTROL with
* CS_STALL Bit set (with No POST_SYNC ENABLED)
*
* Wa_18040903259 says that timestamp are incorrect (not doing the CS Stall
* prior to writing the timestamp) with a command like this:
*/
if (intel_needs_workaround(devinfo, 1607156449) &&
current_pipeline == GPGPU && post_sync_op != NoWrite) {
emit_pipe_control(batch, devinfo, current_pipeline,
NoWrite, ANV_NULL_ADDRESS, 0,
bits, "Wa_1607156449");
bits = ANV_PIPE_CS_STALL_BIT;
}
#endif
#if INTEL_NEEDS_WA_18040903259
/* Wa_18040903259 says that on RCS engine, in GPGPU mode, timestamp are
* incorrect (not doing the CS Stall prior to writing the timestamp) with a
* command like this:
*
* PIPE_CONTROL(CS Stall, Post Sync = Timestamp)
*
@ -2470,13 +2481,13 @@ emit_pipe_control(struct anv_batch *batch,
* first or second PIPE_CONTROL. It seems logical that it should go to the
* first so that the timestamp accounts for all the associated flushes.
*/
if ((intel_needs_workaround(devinfo, 1607156449) ||
intel_needs_workaround(devinfo, 18040903259)) &&
if (intel_needs_workaround(devinfo, 18040903259) &&
batch->engine_class == INTEL_ENGINE_CLASS_RENDER &&
current_pipeline == GPGPU &&
post_sync_op != NoWrite) {
emit_pipe_control(batch, devinfo, current_pipeline,
NoWrite, ANV_NULL_ADDRESS, 0,
bits, "Wa_18040903259/Wa_18040903259");
bits, "Wa_18040903259");
bits = ANV_PIPE_CS_STALL_BIT;
}
#endif