This statement level construct takes no clauses and has no associated
statement, and simply labels a number of array elements as valid for
caching. The implementation here is pretty simple, but it is a touch of
a special case for parsing, so the parsing code reflects that.
This renames the neon_* tests to be more consistent with the original files that they were forked from. This makes it easier for maintainers to see which test files may need to be ported to MSan.
It also fixes the header comment for llvm/test/Instrumentation/MemorySanitizer/AArch64/arm64-vcvt.ll (instrumentation was updated in #126136 aka e9e6ba6a5e2a4ca7386861136196903febb9968b, but the comment was not).
This PR fixes several bugs in `DepthwiseConv2DIsMul`:
- The DepthwiseConv2DOp should restrict the types to integer or float to
prevent a crash.
- `notifyMatchFailure` should be called before creating the new
operations.
Since evaluate::Expr can show up in the parse tree in the semantic
analysis step, make it possible to dump its structure in the Semantics
module.
The Lower module depends on Semantics, so the code is still accessible
in it.
This can be used to trigger implicit module map lookup without also
importing the module. This can be useful for debugging as it avoids
loading the module map from the AST file, which has slightly different
semantics.
The current definition of the CKSM instruction uses GR64 for the first
operand. However, according to the Principles Of Operation the bits 0-31
of the first operand always remain unchanged. This PR changes the first
operand to GR32 to model this.
This has no further implication as this instruction is not used during
code generation.
There was an assert in GetConstant checked if Bound is constant.
However, GetConstant was only called when IsConstant==true.
This refactor attempts to get rid of the assert by combining GetConstant
and IsContstant.
Add AArch64MCSymbolizer that symbolizes `MCInst` operands during
disassembly. The symbolization was previously done in
`BinaryFunction::disassemble()`, but it is also required by
`scanExternalRefs()` for "lite" mode functionality. Hence, similar to
x86, I've implemented the symbolizer interface that uses
`BinaryFunction` relocations to properly create instruction operands. I
expect the result of the disassembly to be identical after the change.
AArch64 disassembler was not calling `tryAddingSymbolicOperand()` for
`MOV` instructions. Fix that. Additionally, the disassembler marks `ldr`
instructions as branches by setting `IsBranch` parameter to true. Ignore
the parameter and rely on `MCPlusBuilder` interface instead.
I've modified `--check-encoding` flag to check symolization of operands
of instructions that have relocations against them.
This PR cleans up cast-overflow.cpp, more specifically:
1. avoid using undefined value as an exit code (old case `9`)
2. narrowing conversions are allowed to produce `inf` and they are
well-defined. Remove dead code (old case `8`)
3. the same applies to the conversion int -> float16. Remove dead code
(old case `7`)
See also
https://clang.llvm.org/docs/UndefinedBehaviorSanitizer.html#:~:text=%2Dfsanitize%3Dfloat%2Dcast,to%20integer%20types.
Currently ubsan doesn't properly detect UB on float16 -> int casts, I
have a fix for that (will send as a separate PR).
(Continuing from #106160)
This PR addresses remaining review comments from the original PR.
Original PR Description
---
Upcoming hardware (gfx12 and some future gfx9) will support the OCP
8-bit float formats for their matrix multiplication intrinsics and
conversion operations, retaining existing opcodes and compiler builtins.
This commit adds support for these types to the MLIR wrappers around
such operations, ensuring that the OCP types aren't used to generate
those builtins on hardware that doesn't expect that format and,
conversely, to ensure that the pre-OCP formats aren't used on new
hardware.
---------
Signed-off-by: Mirza Halilcevic <mirza.halilcevic@amd.com>
Co-authored-by: Paul Fuqua <pf@acm.org>
Co-authored-by: Krzysztof Drewniak <Krzysztof.Drewniak@amd.com>
Previously telemetry was optional at build-time in such a way that none
of it was built at all, but llvm/llvm-project@159b872b37 changed that
and now it's optional in a different way so we need to have it available
in the bazel BUILD graph.
Define the intrinsic `CO_REDUCE` and add semantic checks.
A test was already present but was at `XFAIL`. It has been modified to
take new messages into the output.
Add additional overloads for min and max to support
min(vector<T,N>, T) and min(T, vector<T,N>)
max(vector<T,N>, T) and max(T, vector<T,N>)
Add tests
Closes#128231
The only in-tree user of `bit usePropertiesForAttributes = 0;` was a
series of tests for the output of -gen-op-{decls,defs}. This commit
updates those tests to match the rest of the repository.
In the short term, this is intended to enable testing upcoming updates
to collective builders. In the long term, this is a step in the removal
of usePropertiesForAttributes = 0.
One side effect of these tests updates was the realization that the
autogenerated implementations of removeFooAttr() were not returning the
value of the removed attribute. This issue has been addressed and the
tests have been updated to reflect the change. This is the only
functionality change in this PR.
This patch introduces support for the `DW_AT_LLVM_stmt_sequence`
attribute in the GSYM DWARF transformer. With this change, the DWARF
GSYM creation process can now accurately associate debug information
with the correct functions, even when multiple functions have been
merged together.
The `macho-gsym-merged-callsites-dsym.yaml` test data is regenerated to
include the fixes in the DWARF linker
(https://github.com/llvm/llvm-project/pull/128953) and the test is
updated to check that debug data is correctly associated for merged
functions.
This PR preserve linalg Op types for certain named ops such as Fill,
Copy and Transpose instead of fusion always resulting in a generic Op.
---------
Signed-off-by: Nirvedh Meshram <nirvedh@gmail.com>
Add a pattern that bubbles up tensor.extract_slice through
tensor.expand_shape, and add a transform op to tensor dialect
to directly use this pattern.
This pattern enables tiling and fusing op chains which contain
tensor.expand_shape if added as a cleanup pattern of tile and fuse
utility.
Without this pattern that would not be possible, as
tensor.expand_shape does not implement the tiling interface.
In addition, registering this pattern as a cleanup pattern for
transform.structured.fuse.
The pattern was first implement in IREE project by
Quinn Dawkins and is being upstreamed.
---------
Co-authored-by: Quinn Dawkins <quinn.dawkins@gmail.com>
They are error-prone as MCParser may parse a variant kind,
which cannot be handled by the target.
The replacement in MCAsmInfo should be used instead.
Follow-up to f244b8eed37a12539fb11b76e19ec7a7eb41dccc
The default triple of Amazon Linux on AArch64 is aarch64-amazon-linux,
see issue highlighded by PR #109263, somewhat serious linker issues are
encountered if any other triple is being used.
Unfortunately, this makes XFAIL lines like
`XFAIL: target=aarch64{{.*}}-linux-gnu` ineffective, making it
impossible to complete all of the check-cxx on Amazon Linux without
failing.
We didn't also copy over the next stop hook id, which meant we would
overwrite the stop hooks from the dummy target with stop hooks set after
they are copied over.
Flang generates slower code for `CSHIFT(CSHIFT(PTR(:,:,I),sh1,1),sh2,2)`
pattern in facerec than other compilers. The first CSHIFT can be done
as two memcpy's wrapped in a loop for the second dimension.
This does require creating a temporary array, but it seems to be faster,
than the current hlfir.elemental inlining.
I started with modifying the new index computation in
hlfir.elemental inlining: the new arith.select approach does enable
some vectorization in LLVM, but on x86 it is using gathers/scatters
and does not give much speed-up.
I also experimented with LoopBoundSplitPass
and InductiveRangeCheckElimination for a simple (not chained) CSHIFT
case, but I could not adjust them to split the loop with a condition
on the value of the IV into two loops with disjoint iteration spaces.
I thought if I could do it, I would be able to keep the hlfir.elemental
inlining mostly untouched, and then adjust the hlfir.elemental inlining
heuristics for the facerec case.
Since I was not able to make these pass work for me, I added a special
case inlining for CSHIFT(ARRAY,SH,DIM=1) via hlfir.eval_in_mem.
If ARRAY is not statically known to have the contiguous leading
dimension, there is a dynamic check for contiguity, which allows
exposing it to LLVM and enabling the rewrite of the copy loops
into memcpys. This approach is stepping on the toes of LoopVersioning,
but it is helpful in facerec case.
I measured ~6% speed-up on grace, and ~4% on zen4.
[NVPTX] Add Intrinsics for discard.*
This PR adds intrinsics for all variations of discard.*
* These intrinsics supports generic or global for all variations.
* The lowering is handled from nvvm to nvptx tablegen directly.
* Lit tests are added as part of discard.ll
* The generated PTX is verified with a 12.3 ptxas executable.
* Added docs for these intrinsics in NVPTXUsage.rst.
For more information, refer to the PTX ISA
<https://docs.nvidia.com/cuda/parallel-thread-execution/#data-movement-and-conversion-instructions-discard>_.
---------
Co-authored-by: abmajumder <abmajumder@nvidia.com>
Co-authored-by: gonzalobg <65027571+gonzalobg@users.noreply.github.com>
Simpler detection of dynamic library operands as the readelf one seems
to be unreliable (works on my setup, not on buildbots).
This is a follow-up to #127020
- Trailing whitespace shows up as red on my editor, so remove.
- Docker on my machine warns that having line continuations like:
```
sudo \
foo
```
is deprecated, and will become an error, so fix that up ahead of time.
