fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-07 17:58:26 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 4

intel/fs: Don't use NoDDClk/NoDDClr for split SHUFFLEs

Browse source 
When I copied and pasted the code from MOV_INDIRECT for handling the
dependency controls, I missed a subtle difference between MOV_INDIRECT
and SHUFFLE.  Specifically, MOV_INDIRECT gets lowered to a narrow
instruction on Gen7 by the SIMD width lowering whereas SHUFFLE has to
split it in the generator.  Therefore, the check safety check for
whether or not we can use dependency control has to be based on the
lowered width rather than the width of the original instruction.

Fixes: a8ac61b0ee "intel/fs: NoMask initialize the address..."
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3593
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6989>
(cherry picked from commit 8427e56067)
This commit is contained in:
Jason Ekstrand 2020-10-02 13:37:05 -05:00 • committed by Dylan Baker
parent b7c3713706
commit 6ea01cd07a
2 changed files with 16 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
.pick_status.json
View file

@ -391,7 +391,7 @@
"description": "intel/fs: Don't use NoDDClk/NoDDClr for split SHUFFLEs",
"nominated": true,
"nomination_type": 1,
"resolution": 0,
"resolution": 1,
"master_sha": null,
"because_sha": "a8ac61b0ee2fdf4e8bc7b47aee9c24f96c40435c"
},

18
src/intel/compiler/brw_fs_generator.cpp
View file

@ -652,11 +652,23 @@ fs_generator::generate_shuffle(fs_inst *inst,
uint32_t src_start_offset = src.nr * REG_SIZE + src.subnr;
/* Whether we can use destination dependency control without running
* the risk of a hang if an instruction gets shot down.
/* From the Haswell PRM:
*
* "When a sequence of NoDDChk and NoDDClr are used, the last
* instruction that completes the scoreboard clear must have a
* non-zero execution mask. This means, if any kind of predication
* can change the execution mask or channel enable of the last
* instruction, the optimization must be avoided. This is to
* avoid instructions being shot down the pipeline when no writes
* are required."
*
* Whenever predication is enabled or the instructions being emitted
* aren't the full width, it's possible that it will be run with zero
* channels enabled so we can't use dependency control without
* running the risk of a hang if an instruction gets shot down.
*/
const bool use_dep_ctrl = !inst->predicate &&
inst->exec_size == dispatch_width;
lower_width == dispatch_width;
brw_inst *insn;
/* Due to a hardware bug some platforms (particularly Gen11+) seem