If alpha-to-coverage is exported via mrtz, don't upgrade the mrt0 format
to one with an alpha channel.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20967>
RGP expects a pipeline's shaders to be all stored sequentially, eg:
[vs][ps][gs]
As such, it assumes a single bo is dumped to the .rgp file, with
the following info:
* va of the bo
* offset to each shader inside the bo
For radeonsi, the shaders are stored individually, so we may have
a big gap between the shaders forming a pipeline => we can produce
very large file because the layout in the file must match the one
in memory (see the warning in ac_rgp_file_write_elf_text).
This commit implements a workaround: gfx shaders are re-exported as a
pipeline.
To update the shader address, a new state is created (sqtt_pipeline),
which will overwrite the needed _PGM_LO_* registers.
This reduces DeuxEX rgp captures from 150GB+ to less than 100MB.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18865>
ac_nir_translate can fail now so forward the translation result to
si_llvm_compile_shader. If it's a failure, it'll print something like:
radeonsi: can't compile a main shader part
And the shader won't be used.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18147>
This moves setting those registers from an unconditional place in draw_vbo
into si_set_rasterized_prim (for draw_vbo), si_update_rasterized_prim
(for bind_xx_shader), and si_bind_rs_state.
It's a little more complicated than expected.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18195>
Create nir passthrough shader with explicit input/output and vertex
output count so that it can be handled by compiler same as user tcs.
The drawback is we create more si_shader_selector with different
input/output and vertex output count which was handled by compiler
backend before.
As fixed function tcs can be handled like user tcs, we don't need
the dedicated fixed_func_tcs_shader state either.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16705>
GDS is basically scalar in gfx11.
This is not exactly how it's supposed to be done (we should be using
the GDS_STRMOUT registers), but it works.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16990>
We just need to pass the uses_discard flag to the epilog.
The hw skips the export anyway. This will hang if SPI registers declare
an output format or KILL_ENABLE is set because those cases require
an export with done=1.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16885>
We need to handle the fact that it kills pixels.
v2: also update si_update_ps_inputs_read_or_disabled
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16885>
If the first time the preamble is written, one of the rings
isn't allocated, we wouldn't write the RING_SIZE to the preamble.
Later, when the preamble gets updated after the ring allocation,
the new RING_SIZE packet would overwrite other packets.
To prevent this, always write the RING_SIZE (the alternative would
be to write NOP packets).
This fix "*ERROR* Illegal register access in command stream" hangs
I observed on GFX8.
Fixes: 32c7805ccc ("radeonsi: merge all preamble states into one")
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16962>
Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Acked-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Acked-by: Martin Roukala <martin.roukala@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16604>
The preamble will be skipped by the kernel if there is no context switch.
Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16509>
Tess registers are appended. GS registers are appended or overwritten
if they are already set. There are separate TMZ and non-TMZ preambles.
The preamble will be passed to the kernel as an IB to execute on a context
switch only.
Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16509>
We can't use UINT16_ABGR for the alpha channel. Always use 32_ABGR.
Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16509>
Instead of storing them in a linked list, put them in an array
in si_shader_selector. The keys are also stored separatly, to
avoid pointer chasing when searching a variant in si_shader_select_with_key.
This main point here is to simplify the code by storing everything
in the selector instead of splitting the list storage between the selector
and the shaders; this shouldn't affect performance in a meaningful way.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16273>
This uses the common helper code to implement the tess ring sizing.
One question is if radeonsi should be using tess_offchip_ring_offset
in some places it's using tess_factor_ring_size?
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16415>
