When LLVM 18 is used, pass the no-verify-fixpoint option when
running the instcombine pass. Otherwise LLVM may abort with an
error.
The background here is that this option is enabled by default for
testing purposes, because instcombine is normally only explicitly
invoked like this inside tests. If it is used in an actual
production pipeline, the no-verify-fixpoint option needs to be
enabled.
This should fix the issue reported at
https://bugzilla.redhat.com/show_bug.cgi?id=2268800.
Reviewed-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28101>
This reverts commit ac5d529c65. We're
failing 4 of the tests:
dEQP-VK.reconvergence.subgroup_uniform_control_flow_ballot.compute.nesting4.1.2
dEQP-VK.reconvergence.subgroup_uniform_control_flow_elect.compute.nesting4.1.2
dEQP-VK.reconvergence.workgroup_uniform_control_flow_ballot.compute.nesting4.1.2
dEQP-VK.reconvergence.workgroup_uniform_control_flow_elect.compute.nesting4.1.2
This is because we aren't guaranteeing re-convergence for loop
continues.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28221>
The stride given in the shader is in number of elements of the of the
type pointed by the given pointer, which may not match the matrix own
element type.
Since we cast the pointer to match the element type, the stride needs to
be ajusted accordingly.
v2:
- Fix mismatching bit-width in matrix element type and pointer type (Caio)
- Do the stride calculation in one place
Fixes dEQP-VK.compute.pipeline.cooperative_matrix.khr_*.multicomponent.*
Fixes: 3a35f8b29b ("intel/cmat: Lower cmat_load and cmat_store")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/10820
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27903>
The varying linker isn't changed. The passes are executed after linking
varyings and before linking uniforms if nir->options->lower_io_variables
is true. nir_opt_varyings can move uniforms between shaders and cause them
to be DCE'd.
It requires moving IO deref lowering from st/mesa into the GLSL linker
and nir_opt_varyings should be added at the same time because IO deref
lowering alone would disable IO optimizations in st/mesa such as
compaction.
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26819>
This will make a later commit simpler. No change in behavior.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26819>
Highlights:
- all shader stages and all input/output types are handled, including
inputs and outputs with multiple vertices
- the optimizations performed are: unused input/output removal, constant
and uniform propagation, output deduplication, inter-shader code motion,
and compaction
- constant and uniform propagation and output deduplication work even
if a shader contains multiple stores of the same output, e.g. in GS
- the same optimizations are also performed between output stores and
output loads (for TCS)
- FS inputs are packed agressively. Only flat, interp FP32, and interp
FP16 can't be in the same vec4. Also, if an output value is
non-divergent within a primitive, the corresponding FS input is
opportunistically promoted to flat.
The big comment at the beginning of nir_opt_varyings.c has a detailed
explanation, which is the same as:
https://gitlab.freedesktop.org/mesa/mesa/-/issues/8841
dEQP and GLCTS have incorrect tests that fail with this, see:
https://gitlab.freedesktop.org/mesa/mesa/-/issues/10361
Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26819>
Some of those are no longer needed after moving the sin/cos input fixups
to finalize_nir, while others were are just some unneeded remnants from
nir_to_tgsi era.
Almost no shader-db change on RV530:
total instructions in shared programs: 128940 -> 128939 (<.01%)
instructions in affected programs: 154 -> 153 (-0.65%)
helped: 3
HURT: 2
total cycles in shared programs: 197402 -> 197401 (<.01%)
cycles in affected programs: 263 -> 262 (-0.38%)
helped: 3
HURT: 2
or RV370:
total instructions in shared programs: 83946 -> 83944 (<.01%)
instructions in affected programs: 32 -> 30 (-6.25%)
helped: 2
HURT: 0
total cycles in shared programs: 132829 -> 132827 (<.01%)
cycles in affected programs: 93 -> 91 (-2.15%)
helped: 2
HURT: 0
Signed-off-by: Pavel Ondračka <pavel.ondracka@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28070>
before applying the input range normalization. This allows to move the pass
to finalize nir later without worrying we would apply the fixup twice and
also saves few instructions in wined3d shaders, where d3d9 already
guarantees the correct input range.
RV530 shader-db (and similarly for R4xx) improves few Anno1404 shaders:
total instructions in shared programs: 129040 -> 129022 (-0.01%)
instructions in affected programs: 310 -> 292 (-5.81%)
helped: 5
HURT: 0
no change on RV370
Signed-off-by: Pavel Ondračka <pavel.ondracka@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28070>
Added code guards that isolate DRM code paths so platformss that do not have DRM libraries can compile.
MacOS does not have vf86drm library and headers.
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28130>
On systems where the xcb.h is not in the standard prefixed location,
meson needs to be told where to find it.
Adding the `dep_xcb` to the `libdri`build fixes this.
This issue happened on macOS when the brew package manager was not in the default location.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28130>
This patch allows MacOS to compile the dri[2] systems by switching from direct xf86drm.h to
the pre-existing util/libdrm.h.
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28130>
