Now that they're no longer ralloc'd, we have to be much more careful
about indirects. We have to make sure every time a source or
destination is overwritten, its indirect (if any) is freed. We also
have to choose a memory ownership convention for the rewrite functions.
Assuming that they will be called with the source from some other
instruction, we choose to always make a copy of the indirect (if any).
It's the responsibility of the caller to ensure its copy of the indirect
is freed.
Unfortunately, all this extra logic is going to make
nir_instr_rewrite/move_src/dest more expensive because they now have
all the logic of nir_src/dest_copy instead of a simple struct
assignment. Fortunately, the vast majority of rewrite calls are done by
nir_ssa_def_rewrite_uses which is an SSA-only fast-path.
Fixes: 879a569884 "nir: Switch from ralloc to malloc for NIR instructions."
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12884>
By replacing the 48-byte ralloc header with our exec_node gc_node (16
bytes), runtime of shader-db on my system across this series drops
-4.21738% +/- 1.47757% (n=5).
Inspired by discussion on #5034.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11776>
Right now we're using ralloc to GC our NIR instructions, but ralloc has
significant overhead for its recursive nature so it would be nice to use a
simpler mechanism for GCing instructions.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11776>
We were using the ralloc parent in some places, which should work out to
be the shader I think, but to de-ralloc the instrs we should just pass the
existing shader pointer in.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11776>
This code was being tricky with passing a mem_ctx instead of the shader,
then freeing the mem_ctx when the pass was done and all the parallel
copies had been removed from the shader. Use the right type for instr
creation and do a bit of manual list management to prepare the way for
non-ralloc NIR instrs.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11776>
Calling this lower pass twice in a row would cause spurious
set_vertex_and_primitive_count(0, undef) intrinsics after the proper
set_vertex_and_primitive_count intrinsic. This pretty much turns any
geometry shader into garbage.
Fix this by treating nir_intrinsic_emit_vertex_with_counter and
nir_intrinsic_end_primitive_with_counter just like the non-_with_counter
versions. If no blocks would need set_vertex_and_primitive_count
intrinsics added, exit the pass before doing any work. This prevents
the need for DCE to do extra clean up later.
Since this pass is potentially called multiple times via multiple
invocations of a finalize_nir callback, it is (hypothetically?) possible
that control flow could be changed to add new blocks that need this
intrinsic. The check implemented in this commit should be robust
against that possibility.
v2: Add a_block_needs_set_vertex_and_primitive_count. Suggested by
Timur.
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12802>
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Fixes: e76ae39ae2 ("nir: add support for user defined select control")
Fixes: b56451f82c ("nir: add support for user defined loop control")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12778>
Driver like radeonsi load varying in a scalar manner, so prefer to pack
varying with different interpolation qualifier into same slot to save
space.
But driver like panfrost/bifrost can load varying in vector manner,
so prefer to pack varying with same interpolation qualifier.
Driver can add interpolation qualifiers which are able to be
packed into same varying slot to pack_varying_options nir option.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12537>
nir_foreach_src() bails after cb returns false for any src. Which isn't
the behavior we were looking for. Move progress flag to state struct
instead, so we don't skip visiting some sources.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12732>
Normal UBOs have explicit strides on them, make our lowered one behave the
same.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12175>
Without this, there's no way to match the UBO nir_variable declarations to
the load_ubo intrinsics referencing their data.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12175>
The lowered LS and NGG stages use local_invocation_index and they
can benefit from the unsigned upper bound because they can emit a
less expensive integer multiplication instruction.
This was working in the past, but accidentally borked by a refactor.
Fossil DB changes on Sienna Cichlid:
Totals from 956 (0.74% of 128647) affected shaders:
CodeSize: 2354172 -> 2344712 (-0.40%)
Instrs: 434359 -> 434327 (-0.01%)
Latency: 1883949 -> 1876814 (-0.38%)
InvThroughput: 762638 -> 757405 (-0.69%)
Fossil DB changes on Sienna Cichlid (with NGGC enabled):
Totals from 57873 (44.99% of 128647) affected shaders:
CodeSize: 155844192 -> 155607064 (-0.15%)
Instrs: 29799184 -> 29799152 (-0.00%)
Latency: 130959764 -> 130814224 (-0.11%); split: -0.11%, +0.00%
InvThroughput: 21100300 -> 20928635 (-0.81%); split: -0.81%, +0.00%
Fixes: 8af6766062
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12558>
These won't work since a workgroup can span more than one thread, and
the temporaries are not shared memory.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10600>
Mesh shader outputs are either:
- non-array builtins
- array builtins that are either per-primitive or per-vertex
- user-defined outputs that must be either per-primitive or per-vertex
So we can identify any array output as "arrayed" for the purposes of
I/O lowering.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10600>
Per-primitive is similar to per-vertex attributes, but applies to all
fragments of the primitive without any interpolation involved.
Because they are regular input and outputs, keep track in shader_info
of which I/O is per-primitive so we can distinguish them after deref
lowering. These fields can be used combined with the regular
`inputs_read`, `outputs_written` and `outputs_read`.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10600>
We can't append instructions following a return/halt instruction
because the control flow helpers will modify the successor of the
block containing the return/halt. And the NIR validator enforces that
the return/halt must have the end of the function as successor.
This tends to happen following lower_shader_calls lowering which
inserts halts. This probably doesn't prevent the optimization, it'll
just happen in one of the return shaders after the halt has been
removed.
v2: Move prev block ending check earlier in the function (Daniel)
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12506>
v2: Fix copy-and-paste bugs in lowering patterns.
v3: Add has_sudot_4x8 flag. Requested by Rhys.
v4: Since the names of the opcodes changed from dp4 to dot_4x8, also
change the names of the lowering helpers. Suggested by Jason.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12142>
v2: Add and modify patterns to let constant folding do better.
v3: Remove '(is_not_zero)' from the patterns that try to combine
addends. I honestly don't know why I had it there in the first place,
and nothing in my deep git logs could help clue me in. Noticed by
Alyssa. Remover patterns that detect open-coded udot_4x8. Suggested by
Alyssa and Jason. Add missing sudot_4x8 patterns.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12142>
Six opcodes are added: sdot_4x8_iadd, udot_4x8_uadd, sudot_4x8_iadd,
sdot_4x8_iadd_sat, udot_4x8_uadd_sate, and sudot_4x8_iadd_sat. These
represent the combinations of integer dot-product and add that operate
on packed source vectors. That is, the four 8-bit values for each
vector is stored in a single 32-bit integer.
Some hardware may prefer to operate on unpacked byte vectors. When such
hardware comes to Mesa, we'll have to figure out how to name things.
v2: Add nir_op_iudp4a and nir_op_iudp4a_sat instructions. These opcodes
are not 2-source commutative.
v3: Rename all opcodes to be more like some existing 4x8 opcodes.
Suggested by Timur. Change type of packed vector sources to uint32,
change types of constant folding variables to have explicit size, and
delete some extra casts. All suggested by Jason.
v4: Fix typo previously noticed by Alyssa but missed in v2.
v5: Add has_sudot_4x8 flag. Requested by Rhys.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12142>
Without this, lowered saturating ALU instructions would only clamp to
the range of the new type instead of the range of the old type.
v2: Use nir_iclamp. Suggested by Jason. Use new
u_{int,uint}N_{min,max}() helpers.
Fixes: 090e282407 ("nir: Add a saturated unsigned integer add opcode")
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12142>
This creates a single nir_op_vecn instead of a nir_op_vecn and several
copies.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12469>
