The idea here was that pixel shader framebuffer writes used the g0 and
g1 thread payload register values to construct the message header.
However, most messages are headerless and don't use either. There's a
2012-era comment that the simulator at one point had a bug where certain
headerless messages would incorrectly take the values from the g0/g1
register contents rather than using sideband. But, that was likely
fixed eons ago. So we really don't need to do this.
Furthermore, there are many more shader stages these days:
- VS: r1 contains output URB handles
- TCS: r1 contains ICP handles
- TES: r1 contains gl_TessCoord.x (r4 contains output URB handles)
- GS: r1 contains output URB handles
- CS: r1 contains LocalID.X on DG2+ but nothing on older hardware
- Task/Mesh: r1 contains LocalID.X
- BS: r1 contains bindless stack handles
Vertex and geometry aren't likely to benefit here because r1 is needed
for their output messages, which are also what terminate the shader.
TES will definitely benefit because we were making a value pointlessly
live for the whole program. Same for TCS, to a lesser extent. Compute
prior to DG2 was the worst, as g1 literally has no meaningful content,
so there is no point to keeping it live.
fossil-db on Alchemist shows substantial spill/fill improvements:
Totals:
Instrs: 148782351 -> 148741996 (-0.03%); split: -0.03%, +0.01%
Cycles: 12602907531 -> 12605795191 (+0.02%); split: -0.70%, +0.72%
Subgroup size: 7518608 -> 7518632 (+0.00%)
Send messages: 7341727 -> 7341762 (+0.00%)
Spill count: 54633 -> 52575 (-3.77%)
Fill count: 104694 -> 100680 (-3.83%)
Scratch Memory Size: 3375104 -> 3287040 (-2.61%)
Totals from 301172 (48.21% of 624670) affected shaders:
Instrs: 95531927 -> 95491572 (-0.04%); split: -0.05%, +0.01%
Cycles: 9643531593 -> 9646419253 (+0.03%); split: -0.91%, +0.94%
Subgroup size: 4492512 -> 4492536 (+0.00%)
Send messages: 4399737 -> 4399772 (+0.00%)
Spill count: 20034 -> 17976 (-10.27%)
Fill count: 41530 -> 37516 (-9.67%)
Scratch Memory Size: 1522688 -> 1434624 (-5.78%)
Assassin's Creed Odyssey in particular has 20% fewer fills.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30146>
Doxygen documentation says
> If the file name is omitted (i.e. the line after \file is left
> blank) then the documentation block that contains the \file command will
> belong to the file it is located in.
so we can omit the filename itself when using the annotation.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30168>
Gfx 12.5 moved scratch to a surface and SURFTYPE_SCRATCH has this pitch
restriction:
RENDER_SURFACE_STATE::Surface Pitch
For surfaces of type SURFTYPE_SCRATCH, valid range of pitch is:
[63,262143] -> [64B, 256KB]
The pitch of the surface is the scratch size per thread and the surface
should be large enough to accommodate every physical thread.
So here adding a new field to intel_device_info, setting it in
intel_device_info_init_common() so even offline tools can have it set.
And finally adding a check to fail shader compilation if needed
scratch is larger than supported.
This issue can be reproduced in debug builds when running
dEQP-VK.protected_memory.stack.stacksize_1024 on Gfx 12.5 or newer
platforms.
Ref: BSpec 43862 (r52666)
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30271>
This implements an undocumented workaround for a hardware bug that
affects draw calls with a pixel shader that has 0 push constant cycles
when TBIMR is enabled, which has been seen to lead to a hang with
Fallout 3 and Metal Gear Rising Revengeance. This hardware bug has
been reported as HSDES#22020184996 which is still pending a resolution
by the hardware team. However since this workaround found empirically
has been confirmed to fix the issue reliably and it's relatively
harmless it seems worth checking in already even though no final W/A
number is available nor has the W/A json file been updated.
To avoid the issue we simply pad the push constant payload to be at
least 1 register. This is enabled via a brw_wm_prog_key since the
driver needs to be in agreement with the compiler on whether the dummy
push constant cycle is present, and it can be avoided in cases where
the driver knows that TBIMR will be disabled (e.g. for BLORP).
Related: https://gitlab.freedesktop.org/mesa/mesa/-/issues/10728
Related: https://gitlab.freedesktop.org/mesa/mesa/-/issues/11399
Fixes: 57decad976 ("intel/xehp: Enable TBIMR by default.")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30031>
Instead of having a hardcoded list of endian-independent format aliases
in the header, generate them from the format definitions.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29649>
If you have something like this :
con 32 %66 = @load_reg (%62) (base=0, legacy_fabs=0, legacy_fneg=0)
con 32 %27 = @resource_intel (%22 (0xdeaddead), %66, %67, %17 (0x0)) (desc_set=2, binding=96, resource_intel=0, resource_block_intel=-1)
Just copying the brw_reg in ssa_values[] is not enough for the
load_reg intrinsic. We need to call get_nir_src() to force some logic
to create the register correct.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: b8209d69ff ("intel/fs: Add support for new-style registers")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30050>
I haven't been able to find this restriction mentioned anywhere in the
hardware documentation, but the simulator has code to reject this case
as invalid, and it doesn't appear to work on hardware anymore.
Having lower_regioning() handle this takes care of the issue so we
don't have to worry about generating it in random places.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/11489
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30140>
We were generating odd instructions like:
math inv(8) g93<1>HF g85<8,8,1>HF null<8,8,1>F { align1 1Q @7 $4 };
It's unclear whether the type of the null operand matters, but sometimes
these things don't get ignored properly. Out of caution, retype the
null source to match the actual operand's type. It'll at least look
less surprising in assembly dumps.
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30193>
Handle '\n' when inside the MSGDESC start condition,
otherwise the lexer would apply its default rule (write
to stdout).
Without that, newlines were "leaking" to the output when
parsing a multiple line "MsgDesc". E.g. given the file
example.asm below
```
send(8) nullUD g126UD nullUD 0x02000000 0x00000000
thread_spawner MsgDesc: mlen 1 ex_mlen 0 rlen 0
{ align1 WE_all 1Q @1 EOT };
```
the assembler would produce one extra newline
```
$ brw_asm -t hex -g tgl example.asm
31 01 03 80 04 00 00 00 0c 7e 00 70 00 00 00 00
```
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30100>
The function handles the Xe2 case where NibCtrl is gone. Also add
error messages for invalid input when assembling for Xe2, e.g. "2N".
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30060>
Commit f695a9fed2 moved the 64-bit float <-> 16-bit float conversion
splitting into a core NIR pass, so the code remaining here is only
needed for 64-bit integer types.
Presumably in an attempt to remove the float handling, it replaced
simple bit_size == 64 checks with this expression:
(full_type & (nir_type_int64 | nir_type_uint64))
I believe that the intended expression was:
(full_type == nir_type_int64 || full_type == nir_type_uint64)
Unfortunately, the former is incorrect. Any integer or unsigned
NIR type would trigger the former expression. For example:
nir_type_uint32 & (nir_type_int64 | nir_type_uint64) => nir_type_uint
This meant that we were splitting e.g. u2f16 on 32-bit unsigned types
into u2f32 and f2f16, when we can easily natively handle that case.
To fix this, we go back to simple bit_size == 64 checks. This pass is
already run after nir_lower_fp16_casts which will split the float case,
so we will never see it here.
fossil-db on Alchemist shows a -1.14% reduction in affected shaders for
google-meet-clvk shaders. In another ChromeOS workload, it improves
performance by around 8% on Meteorlake.
Thanks to Sushma Venkatesh Reddy for finding this performance issue!
Fixes: f695a9fed2 ("intel/compiler: use nir_lower_fp16_casts")
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Ivan Briano <ivan.briano@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30091>
Indirect addressing(vx1 and vxh) not supported with UB/B datatype for
src0, so we need to change the data type for both dest and src0.
This fixes following tests cases on Xe2+
- dEQP-VK.spirv_assembly.instruction.compute.8bit_storage.push_constant_8_to_16*
- dEQP-VK.spirv_assembly.instruction.compute.8bit_storage.push_constant_8_to_32*
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29316>
We can do CSEL on F, HF, *W, and *D on Gfx11+. Gfx9 can only do F.
We can lower unsupported types to CMP+CSEL, allowing us to use CSEL
in the IR and not worry about the limitations.
Rework: (Sagar)
- Update validation pass for CSEL
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29316>
Reorganize the code to make clearer all the lowering cases:
(a) Single invocation workgroup. Index and IDs are all zero.
(b) Local ID provided by hardware.
(c) Local Index provided by the hardware. Depending on the case this
might not be the final local index, e.g. heuristics for tile.
(d) Neither provided by the hardware.
Case (c) is new and supported by Mesh/Task shaders. At the moment the
nir_lower_compute_system_values handle lowering of LocalID for
Task/Mesh, but a later patch will flip that on ANV.
This will make the Task/Mesh use the same lowering as Compute shaders.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29828>
