fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2025-12-20 22:30:12 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

intel/fs: Always stall between the fences on Gen11+

Browse source 
Be conservative in Gfx11+ and always stall in a fence.  Since there are
two different fences, and shader might want to synchronize between them.

This change also brings back the original code block for the stall
between the fence and comment from the commit
b390ff3517.

v2: (Caio)
 - Re-arrange code block.
 - Adjust comment.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6958

Fixes: f7262462 ("intel/fs: Rework fence handling in brw_fs_nir.cpp")
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Tested-by: Mark Janes <markjanes@swizzler.org>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20996>
This commit is contained in:
Sagar Ghuge 2023-01-30 10:41:37 -08:00 • committed by Marge Bot
parent 51ea81c0a1
commit 0c083d29a5
1 changed files with 13 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

15
src/intel/compiler/brw_fs_nir.cpp
View file

@ -4584,6 +4584,15 @@ fs_visitor::nir_emit_intrinsic(const fs_builder &bld, nir_intrinsic_instr *instr
assert(fence_regs_count <= ARRAY_SIZE(fence_regs));
/* Be conservative in Gen11+ and always stall in a fence. Since
* there are two different fences, and shader might want to
* synchronize between them.
*
* TODO: Use scope and visibility information for the barriers from NIR
* to make a better decision on whether we need to stall.
*/
bool force_stall = devinfo->ver >= 11;
/* There are four cases where we want to insert a stall:
*
* 1. If we're a nir_intrinsic_end_invocation_interlock. This is
@ -4599,10 +4608,12 @@ fs_visitor::nir_emit_intrinsic(const fs_builder &bld, nir_intrinsic_instr *instr
* scheduling barrier to keep the compiler from moving things
* around in an invalid way.
*
* 4. On platforms with LSC.
* 4. On Gen11+ and platforms with LSC, we have multiple fence types,
* without further information about the fence, we need to force a
* stall.
*/
if (instr->intrinsic == nir_intrinsic_end_invocation_interlock ||
fence_regs_count != 1 || devinfo->has_lsc) {
fence_regs_count != 1 || devinfo->has_lsc || force_stall) {
ubld.exec_all().group(1, 0).emit(
FS_OPCODE_SCHEDULING_FENCE, ubld.null_reg_ud(),
fence_regs, fence_regs_count);