19806cc9e4
nak: Initialize RA with only live registers
...
While coping the entire RA struct from the predecessor is probably
faster, it may contain values which are not live in the current block.
We could purge those but the iteration gets tricky with Rust. It's far
clearer and more rust-friendly to deal with it as an initialization
problem.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
ec47843cca
nak: Record register pressure in liveness
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
7c98dd8bfd
nak: Add a PerRegFile helper struct
...
Sometimes it's useful to have something that's per-register-file like we
do for register allocation. Since this is generally useful, add a
struct for it. In future, it might be neat to pull in the enum_map
crate which basically does this generically.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
b661279e6e
nak: Rework liveness to add next-use information
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
f67fecf196
nak: Compute liveness in reverse block order
...
If we walk the blocks forwards, we end up having a minimum O(n^2)
algorithm because everything has to propagate bottom to top. Looping
over the blocks bacwards ensures that all the liveness information is
propgated in the first pass in the absence of back edges.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
82f0c6cf77
nak: Use the builder in some lowering passes
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
2d69a2c1d7
nak: Add a builder
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
063c9f41fa
nak: Work in terms of bits for type sizes
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
ced6b44ba6
nak: Follow memcpy semantics with OpParCopy
...
Destination first followed by source. Otherwise, we'll screw ourselves
up endlessly.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
831d1ee2d6
nak: Do not allocate vectors needlessly in optimization passes
...
All passes allocate a Vec per Instr during map(). This is wasteful,
because most instances of map() produce a single instruction (by
mapping one instruction to another instruction) or no instructions at
all.
In such cases, they return an empty Vec, or a Vec with a single entry.
Rework the signatures so that a Vec is only when mapping one instruction
into many.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
5014b4697d
nak: Heap-allocate Instrs
...
Heap-allocate Instrs to avoid copying them around whenever they are
mutated by a pass. This lowers the amount of copies in detriment of
cache-locality.
Signed-off-by: Daniel Almeida <daniel.almeida@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
fa0891d37c
nak: Implement nir_op_b2b1 and nir_op_b2b32
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
3bf0882ea8
nak: Implement indirect UBO loads
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
0ff7fbf093
nak: Rework cbufs a bit
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
86684fa9aa
nak: Legalize everything
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
9fcc0eaa8a
nak: Fix multisampled textureing
...
Needed to handle txf_ms a few places and force LOD_MODE_ZERO
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
cc88a1c78e
nak: Add a Pred struct move the enum to PredRef
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
38c9d6dfbb
nak: Add accum predicates to Op[FI]Setp
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
c6edf43f70
nak: Delete unused imports and dead code
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
d48bea6638
nak: Fold fsat into FAdd/FFma/FMul
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
4c3a5b16e9
nak: Rework source modifiers instructions a bit
...
Instead of I/D/FMov instructions, just use [DF]Add instead. For ineg, we
add a new INeg instruction which we can lower to IADD3 later. The
reason for this is that IAdd3 is complicated and makes detecting an ineg
rather annoying. Also, if we ever bring NAK up on older hardware, not
all hardware has IAdd3 and INeg will be lowerable everywhere.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
8ba1429216
nak: DCE things with constant false predicates
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
818ec3242b
nak: Optimize OpLop3 and OpPLop3
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
59f6d657f3
nak: Fix the 2nd predicate on LOP3
...
I put it in the wrong spot. Also, it's a source so we should use
set_pred_src() for it.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
4367e0786b
nak: Implement image atomics
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
82e5b0dd93
nak: Implement global/shared_atomic_comp_swap
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
d359c64b2a
nak: Implement nir_intrinsic_shared_atomic_*
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
d1c56b12b9
nak: Implement nir_intrinsic_global_atomic_*
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
18a11aa449
nak: Rework copy-prop to use soruce type decorations
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
9bc2bdd78f
nak: Add a Src::supports_src_type() helper
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
42d31b4bfd
nak: Rework source modifiers a bit
...
They now carry some type informaiton so FNeg and INeg are different
modifiers. Since only INeg exists in hardware, we don't need IAbs or
INegAbs.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
a2390fdbc1
nak: Only divide FS inputs by .w for smooth interpolation
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
7398a262da
nak: Decorate sources with types
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
4def64545a
nak: Add a mechanism for decorating sources with types
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
3207f29a09
nak: Integers don't have abs() source modifiers
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
2a2103d73b
nak: Implement load/store_shared
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
73976c3f31
nvk: Plumb num_barriers through from NAK
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
4c2ce288f9
nak: Implement barriers
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
53a372ec3d
nak: Add F2F and implement fquantize16
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
dc3b2c7d10
nak: sin() and cos() require we divide by 2pi
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
a6e1167a83
nak: Don't allow r255 in texture or surface ops
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
b70d998180
nak: compiler: replace Instr::new(..) with OpFoo {}.into()
...
As Instr now implements From<T: Into<Op>>, replace
Instr::new(Op::Foo(OpFoo {})) with OpFoo {}.into() to improve
readability.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
ab9cf27c02
nak: compiler: add From<T:Into<Op>> for Instr
...
Adding this From implementation makes it possible to clean up
the code in the next commit.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
e8f220f7a1
nak: Legalize Op[FI]Setp
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
3bd42dd254
nak: Improve [FI]SETP encoding
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
3ac03b043e
nak/nir: Lower image size and samples to txq
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:10 +00:00
6750d21e65
nak: Legalize more stuff
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:09 +00:00
854b4c0817
nak: Only put actually live SSA values in the ra.live_in sets
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:09 +00:00
49f8fb0c03
nak: Legalize OpShf
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:09 +00:00
4e4548afc9
nak: Improve RA failure messages
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24998 >
2023-11-14 00:48:09 +00:00
Faith Ekstrand