Generates bi_index from nir_alu_src, taking into account the applied
swizzle, and using (swizzle / 32-bit) portion as an offset, to be
applied later during RA. The sub 32-bit portion only applies for 8-bit
and 16-bit instructions, which need to either handle them explicitly as
a swizzle specifier, or lower to a swizzle explicitly.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
There's no flag/mode to reflect when a LOD is zero on FETCH instructions,
we have to emit the LOD/CUBE word unconditionally.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7653>
If we don't add this check we clobber fetch mode when a 0 LOD is
specified on a txf instruction.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7653>
v2: Drop register tracking since it was wrong, and meaningful accounting
is tricky for Bifrost (which wants round robin RA for at least some
registers)... we'll cross that bridge when we get there, possibly
preferring a "max liveness" estimate to the raw RA output.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7615>
Prevents attempts to spill tied TEXC src/dest, resulting in a crash,
seen in shaders/tesseract/229.shader_test
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7615>
We need that to account for potential swizzling on the source reg.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7615>
Enough to get shaders/humus-celshading/1.shader_test compiling, no idea
if it actually works but it looks reasonable.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7615>
src0 and src1 were mixed leading to invalid varying indices. In order to
fix that properly, we first extend load_vary to pass the immediate index
through a dedicated field and add a special boolean. This way, we don't
have to make sure src0 always contains the index, and can instead match
the src numbering defined in ISA.xml.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7636>
It's just clearer to have dedicated fields encoding the fact that the
LD_VAR should be SPECIAL, and another field storing the special var id.
With this change, the source index know matches the ISA.xml definition.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7636>
Let the compiler pass the update mode instead of inferring from the
constant value.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7636>
Some instructions differ between v6 and v7 and we'll need to know which
architecture we're compiling for if we want to generate the right
instructions.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7530>
Now that we lower uniforms to UBO we can get rid of bi_emit_ld_uniform().
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7472>
The number of src swizzle to initialize depends on the number of source
properties (size and number of components) not the destination ones.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7472>
So we can extend bi_emit_ld_vary() to support centroid and sample modes.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7472>
We need to do the transform specified in the OpenGL spec ourselves, with
some assistance from the hardware.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7408>
Some special instructions are scheduled on the FMA unit, let's add a
new class for this case and rename the old one accordingly.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7408>
nir_tex_instr_src_index() returns a negative result when the requested
source does not exist, but we cast that to an unsigned type thus losing
this information.
Fixes: b83c293674 ("pan/bi: Add basic support for txf_ms")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7314>
We don't handle undefs explicitly in NIR->BIR which means if they aren't
optimized out they won't be RA'd to anything and then backend RA will
crash (as occurs in a glamor shader seen in MATE).
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7206>
Letting the caller zero-initialize the program object is error prone,
not to mention that resources attached to the program might not be freed
by the caller. Let's simplify that by letting the compiler allocate the
panfrost_program object. Those objects should be freed with ralloc_free().
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7206>
We currently don't support txf_ms instructions specifying a texel offset
src.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7206>
No need to add a COMBINE instruction if TEXC only needs zero or one
staging reg.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7206>
Blend shaders are passed blend inputs through r0-r3. Let's emit a MOV
from those register when we see a load_input intrinsic.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7151>
Blend shaders shouldn't use the blend descriptors stored in the FAU RAM
since this is what triggered the blend shader call in the first place.
The descriptor is instead extracted from the compiler inputs and passed
as a constant to the blend instruction.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7151>
We will need that for blend shaders so they can be passed a return
address and jump back to the fragment shader when they're done.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7151>
This is mapped to the LD_TILE instruction. Note that multi-sample RTs
are not supported yet.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7151>
The current logic assumes blend descriptors are always retrieved from
the blend descriptor slots present in the FAU RAM, but this assumption
no longer stands when we add blend shaders to the mix. In that case we
need to use an 'opaque blend' whose descriptor is passed through
embedded constants.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7151>
The uniform_constant field and BIFROST_SRC_CONST_{LO,HI} definitions
seem to imply that those only deal with embedded constants. Let's
rename them to reflect the fact that they actually encode accesses to
the Fast-Access-Uniform RAM.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7151>
For texelFetch. A few earlier header fields were wrong.
Fixes dEQP-GLES2.functional.texture.mipmap.2d.generate.rgba8888_fastest
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7081>
There's a different LOD format used for TXF that will be able to
accomodate cube maps as well.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7081>
Also uses the 8.8 signed fixed-point format.
dEQP-GLES2.functional.shaders.texture_functions.fragment.texture2d_bias
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7081>
We need to emit some lowering code to convert it to the 8:8 format
expected by the hardware.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7081>
Lower to a COMBINE, which in turn will lower to moves so RA does the
right thing.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7081>
We still need to handle actual complex texturing instructions, as well
as packing, but this is the start.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7081>
And fix the bad assertion that let this slip.
Like combines, nir_op_vec can be vector, and we need to lower this
ourselves. Thankfully, the lowering is simple.
Fixes
dEQP-GLES2.functional.shaders.loops.for_uniform_iterations.nested_tricky_dataflow_1_*
Fixes: b2c6cf2b6d ("pan/bi: Eliminate writemasks in the IR")
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7081>
If a block ends in an unconditional jump, we don't need to record a
fallthrough successor as well, since it's unreachable.
Likely harmless but makes the IR harder to parse with messy CF graphs.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7081>
So we can extend it more easily without having to patch all callers.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7066>
Varying stores require us to generate a LEA_ATTR_IMM instruction, which
is a load.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Fixes: 0aa08ae2f6 ("nir: Split NIR_INTRINSIC_TYPE into separate src/dest indices")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6985>
We're about to introduce conversion ops which are going to want two
different types. We may as well just split the one we have rather than
end up with three. There are a couple places where this is mildly
inconvenient but most of the time I find it to actually be nicer.
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6945>
This makes it explicit that this intrinsic is only for SSBOs. For the
v3dv driver, we'll be adding a get_ubo_size intrinsic and we want to be
able to distinguish between the two.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6812>
This maps more closely to the hardware, which makes for easier packing.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>
Logically, it doesn't matter, but we want the IR to accurately reflect
the hardware behaviour.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>
To make register_format packing explicit, and possibly in the future
support the auto mode.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>
Needed to select between global, UBO, TLS, and WLS addressing modes,
required to implement loads/stores correctly.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>
On FMA, a carry/borrow is required for iaddc/isubb (whereas the ADD
counterparts don't support carrying/borrowing). The trick is to model
this with an extra dummy (ZERO) argument which is free to encode on FMA,
and in the scheduler, "demote" to the non-carried versions if we want to
schedule to ADD.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6749>
This also fixes the inverted last parameter of nir_lower_flrp in most drivers.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6599>
It was always fneu but naming it fne causes confusion from time to time. So
lets rename it. Later we also want to add other unordered and fne, this is
a smaller preparation for that.
Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6377>
8.74KB leaked over 52 calls from:
0xffffbb5b9fc3
in ??
_mesa_hash_table_init
at ../src/util/hash_table.c:163
in /home/alyssa/rockchip_dri.so
_mesa_hash_table_create
at ../src/util/hash_table.c:186
_mesa_hash_table_u64_create
at ../src/util/hash_table.c:701
in /home/alyssa/rockchip_dri.so
panfrost_nir_assign_sysvals
at ../src/panfrost/util/pan_sysval.c:130
in /home/alyssa/rockchip_dri.so
midgard_compile_shader_nir
at ../src/panfrost/midgard/midgard_compile.c:2905
in /home/alyssa/rockchip_dri.so
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Cc: mesa-stable
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6373>
Unfortunately this doesn't map nicely to the existing instruction
classes, so we'll make a new one for now.
Signed-off-by: Chris Forbes <chrisforbes@google.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6091>
Bifrost's bitwise ops include the shift capability. Previously we had
hardcoded the shift to zero in all cases.
There's room in future to emit slightly better code if a shift and a
bitwise operation can be folded together, but not going after that for
now.
This change also removes the separate BI_SHIFT instruction class as
BI_BITWISE can cover both cases.
Signed-off-by: Chris Forbes <chrisforbes@google.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6091>
Gallium drivers should never see nir_var_uniform because gallium lowers
regular uniforms to a UBO. No GL driver should ever see either
nir_var_mem_shared because that's lowered in GLSL IR.
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5418>
At the NIR level this is a second vector source of the first (only)
argument; at the BIR level this is a pair of scalars.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5307>
We need proper scheduling for 4-src ops to work, so for now disable
condition fusing so we cap at 3-src at a performance penalty. A bit of a
hack but I'd rather not build hacks into a scheduler that will be
rewritten soon anyway.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4883>
Since the hardware doesn't support them, they're a burden to deal with,
so let's ensure we never get to a place where we would need to at all.
Disables COMBINE lowering for now.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4766>
This moves the fi_types to a new mesa_private.h and removes the
imports.c file. The vast majority of this patch is just removing
pound includes of imports.h and fixing up the recursive includes.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3024>
These require a lot more handholding in the IR than we can deal with at
this stage; we need to restrict ourselves to frcp/sqrt.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4470>
It turns out ports need to be in order of the arguments of an
instruction (port 3, that is), which breaks on instructions whose IR
argument order is different from the packed order, like csel. So fix
that.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4458>
We add a zero argument, we want it to align with the size of whatever
the other arguments were for optimization.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4382>
This way we always have regular csel conditions instead of a weird
.always special case for 3-src CSEL mode.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4276>
When generating csel instructions, we can peak to see what condition is
being used. If we're using a "nice" condition, we can fuse it in with
the csel itself, ideally letting the condition itself be DCE'd away.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4276>
We would like to use this routine opportunistically when fusing
conditions into csels and branches, so let's add a mode where we don't
abort.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4276>
We need to make the semantics of BI_VECTOR a bit more precise -
vectorize only the first argument, not all of them. This is enough for
current and future users, as far as I know.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4276>
We need ALU mostly scalarized, but we get vector moves created from
lower_vec_to_mov so let's scalarize that ourselves rather than bother
NIR.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4276>
We split off MOV from FMOV since the canonical move on Bifrost doesn't
accept modifiers. (We can still do fmov, but with something like add-0.)
This will also make copyprop a little nicer, I think. Anyway, the
non-modifier version we can implement as-is for FMA.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4276>
ATEST actually takes two sources and has a destination. Although the
details are a little funny, we should still model this correctly.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4242>
For direct, this is just 0, but for indirct, this is a sample mask
preloaded in R61. Handle this at code emit time instead of trying to do
crazy monkeypatching later.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4242>
At the moment, we just iterate the clauses in the post-RA, post-sched IR
and generate a dummy clause corresponding, passing the results to the
disassembler to verify.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4242>
Once LCRA has run, we have a map from IR indices to byte offsets into
the register file, so we need to "install" these results, rewriting the
IR to use native registers and fixing up writemasks/swizzles to
substitute vectorization for adjacent registers (for LCRA, we're
modeling in terms of real vectors).
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4158>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4158>
We model the machine as vector (with restrictions) to natively handle
mixed types and I/O and other goodies. We use LCRA for the heavylifting.
This commit adds only the modeling to feed into LCRA and spit LCRA
solutions back; next commit will integrate it with the IR.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4158>
We want types to be consistent throughout the IR so we don't have to
make exceptions to parse things out. These cases just got missed.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4158>
We have native FMA which works for graphics usage (unlike Midgard where
it's really reserved for compute for various reasons), let's use it.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4150>
We can move e v e n more code to be shared and let bi_block inherit from
pan_block, which will allow us to use the shared data flow analysis.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4150>
Now that it's all abstracted nicely with an implementation shared with
Midgard, this is pretty easy to get.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4150>
Only on supported GPUs at the moment; for older Bifrost that don't
support these, I'm not sure yet where the right place to do the lowering
is. NIR algebraic rules would be "nice" but probably impractical -- but
it wouldn't be hard to do it directly in BIR (as a lowering pass or
alternative implementation).
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4139>
These are all regular ALU ops found in GLES2 which makes them
particularly nice targets at the moment. Just translate straight to our
IR.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4139>
Soon we'll have a NIR support to handle this the Right Way along with
pos and sat_signed support, but we'll always need the fallback anyway.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4139>
This will allow us to optimize out the constant moves (although that
will require a DCE pass which has yet to be written).
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4139>
We do the bare minimum translation, just enough for fmov/fadd/fmul right
now with no modifiers / inlined constants / etc. The rest is to come!
But hopefully I got bitsize handling right this time around.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4139>
I feel so dirty. But this will let the IR be a lot more flexible seeing
as we really are vector in a certain sense (I/O, small types)
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4139>
We'd really rather not emit extracts. We are approaching on a vector IR
anyway which is annoying but really necessary to handle I/O and fp16
correctly. So let's just go all the way and deal with swizzles and masks
within reason; it'll still be somewhat saner in the long-term.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4139>
Do the absolute simplest possible thing -- create a clause for every
instruction, and just pick whichever slot we can, nopping the other,
copying whatever constant we have whether it's used or not.
To be clear - this is not to be used in a production compiler. But this
lets actual bundles and clauses show up in the BIR, which unblocks work
on final code generation and packing (which can happen more or less in
parallel to NIR->BIR, optimization, register allocation, and writing an
actual scheduling).
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4097>
In the laziest possible way... We can just emit worst case moves which
DCE will eat for breakfast anyway, and inline constants on instructions
where that is supported directly. This approach eliminates a lot of
nasty corner cases that Midgard's crazy cache scheme has hit, at the
expense of slightly more work for DCE (but it's only a single iteration
of an O(N) pass that has to run anyway..)
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4097>
Not very useful without also handling breaks and continues, of course.
We use the strategy from v3d (vir_to_nir) instead of Midgard's, since
the latter is mildly insane. I mean, it passes deqp but...
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4097>
Copypaste from Midgard with some cleanups. That seems to be a trend
these days. Hopefully boilerplate will come to a close soon.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4097>
Pretty much a copypaste from Midgard except where architectural
decisions diverge around vectorization. On that note, we will need our
own ALU scalarization pass at some point (or rather we'll need to extend
nir_lower_alu_scalar) to allow partial lowering for 8/16-bit ops. I.e.
we'll approximately need to lower
vec4 16 ssa_2 = fadd ssa_0, ssa_1
to
vec2 16 ssa_2 = fadd ssa_0.xy, ssa_1.xy
vec2 16 ssa_3 = fadd ssa_0.zw, ssa_1.zw
vec4 16 ssa_4 = vec4 ssa_2.x, ssa_2.y, ssa_3.x, ssa_4
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4097>
Modeled after the Midgard system. Already we know of two
compiler-visible differences between G52 and G71, so let's keep track so
we can eventually port the compiler to other Bifrost systems.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4061>
We're making some pretty dramatic design pivots so this early on it'll
be easier to start from scratch, I think.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4061>
Currently the schedule_program implementation being used is picked
at compile time, which on the Android platform means that the
bifrost compiler & scheduler is used for all targets, including
midgard based hardware.
This commit disambiguates between the two schedule_program functions.
Signed-off-by: Robert Foss <robert.foss@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Set of opcodes doesn't have enough flexibility in certain cases. E.g.
Utgard PP has vector conditional select operation, but condition is always
scalar. Lowering all the vector selects to scalar increases instruction
number, so we need a way to filter only those ops that can't be handled
in hardware.
Reviewed-by: Qiang Yu <yuq825@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Signed-off-by: Vasily Khoruzhick <anarsoul@gmail.com>
Panfrost is the only user of the macro; we are better off expanding than
having random stuff in nir.h.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Alyssa Rosenzweig
Boris Brezillon
Jesse Natalie
Jason Ekstrand
Marek Olšák
Karol Herbst
Rhys Perry
Chris Forbes
Tomeu Vizoso
Dylan Baker
Icecream95
Robert Foss
Vasily Khoruzhick