Introduce a new Transform dialect extension that uses IRDL op
definitions as matcher descriptors. IRDL allows one to essentially
define additional op constraits to be verified and, unlike PDL, does not
assume rewriting will happen. Leverage IRDL verification capability to
filter out ops that match an IRDL definition without actually
registering the corresponding operation with the system.
Rather than force callbacks for outgoing requests to parse the result
JSON themselves (of type `llvm::Expected<llvm::json::Value>`), allow
users to specify the result type, which
`MessageHandler::outgoingRequest` will parse for them. This eliminates
boilerplate for users sending outgoing requests.
This PR builds on https://github.com/llvm/llvm-project/pull/79494 with an additional path for efficient unsigned `i4 ->i8` type extension for 1D/2D operations. This will impact any i4 -> i8/i16/i32/i64 unsigned extensions as well as sitofp i4 -> f8/f16/f32/f64.
This test shows a few cases (not at all complete) where the current
ArmSME tile allocator produces incorrect results. The plan is to resolve
these issues with a future tile allocator that uses liveness
information.
Previously if the producer tensor.unpack op had "unpadding" semantics,
the folding pattern would construct a destination that does not match
with the result type of the transpose. Because both ops are DPS we can
just reuse the destination of the transpose.
Additionally cleans up a bunch of trailing whitespace in the test file.
Expand `arith.minsi`, `arith.minui`, `arith.maxsi`, `arith.maxui` into
`arith.cmpi` and `arith.select`.
---------
Co-authored-by: Jakub Kuderski <kubakuderski@gmail.com>
This patch generalizes tensor.expand_shape and memref.expand_shape to
consume the output shape as a list of SSA values. This enables us to
implement generic reshape operations with dynamic shapes using
collapse_shape/expand_shape pairs.
The output_shape input to expand_shape follows the static/dynamic
representation that's also used in `tensor.extract_slice`.
Differential Revision: https://reviews.llvm.org/D140821
---------
Signed-off-by: Gaurav Shukla<gaurav.shukla@amd.com>
Signed-off-by: Gaurav Shukla <gaurav.shukla@amd.com>
Co-authored-by: Ramiro Leal-Cavazos <ramiroleal050@gmail.com>
Adds an example that combines loop peeling and scalable vectorisation of
`linalg.depthwise_conv_2d_nhwc_hwc`. This is similar to
transform-op-peel-and-vectorize.mlir and is meant to demonstrate how to
avoid masking when vectorising using scalable vectors.
For all means and purposes llvm.mlir.addressof acts like a constant, and
should be treated as such by passes. In particular, the operation should
be propagated rather than passed whenever possible.
Extends `omp.private` with a new region: `dealloc` where deallocation
logic for Fortran deallocatables will be outlined (this will happen in
later PRs).
This ensures the explicit value is generated (and not a load into the
values array). Note that actually not storing values array at all is
still TBD, this is just the very first step.
Add support for outgoing requests to `lsp::MessageHandler`. Much like
`MessageHandler::outgoingNotification`, this allows for the message
handler to send outgoing messages via its JSON transport, but in this
case, those messages are requests, not notifications.
Requests receive responses (also referred to as "replies" in
`MLIRLspServerSupportLib`). These were previously unsupported, and
`lsp::MessageHandler` would log an error each time it processed a JSON
message that appeared to be a response (something with an "id" field,
but no "method" field). However, the `outgoingRequest` method now
handles response callbacks: an outgoing request with a given ID is set
up such that a callback function is invoked when a response with that ID
is received.
This reverts the revert commit 6844c2feae, which was comprised
of the following commits:
1. f3f6f22dfc - [mlir-lsp] Initialize `Reply::method` member (#89857)
2. 37e13d4924 - [mlir-lsp] Log invalid notification params (#89856)
3. ba1b52e6e7 - [mlir-lsp] Add `outgoingNotification` unit test
4. 84bc21f910 - [mlir-lsp] Add transport unit tests (#89855)
Of these, (4) specifically caused issues in Windows pre-merge buildbots,
in the `TransportTest.MethodNotFound` unit test that it added. The
failure was caused by a statement that asserted that opening a file
stream on a newly created temporary file did not result in an error, but
this assert failed on Windows.
This patch adds additional error logging for failures, to make it
clearer what went wrong when failures occur. This patch also addresses
the Windows failure by ensuring temporary files are created in the
system temporary directory.
Adding `erf` as unary and `powf` as binary.
Same as `max(arg, 0.0)` for `ReLU`, `powf(arg, const)` can be either a
generic (with broadcast) or a pair (`linalg.broadcast + linalg.powf`)
and then lowered "correctly". Either way, the lower dialects need to
know what kind of broadcast anyway, so no materialization of the
constant tensors should remain.
I want to flush the easy ones before we start working on type cast &
softmax.
This patch is moving out following intrinsics:
* vector.interleave2/deinterleave2
* vector.reverse
* vector.splice
from the experimental namespace.
All these intrinsics exist in LLVM for more than a year now, and are
widely used, so should not be considered as experimental.
Adds a few basic polynomial ops.
- add, sub, mul
- mul_scalar
- leading_term, monomial_mul, monomial (useful for lowering `mul` to
standard MLIR)
- from_tensor, to_tensor, constant
---------
Co-authored-by: Jeremy Kun <j2kun@users.noreply.github.com>
Co-authored-by: Oleksandr "Alex" Zinenko <ftynse@gmail.com>
…ted. (#89998)" (#90250)
This partially reverts commit 7aedd7dc75.
This change removes calls to the deprecated member functions. It does
not mark the functions deprecated yet and does not disable the
deprecation warning in TypeSwitch. This seems to cause problems with
MSVC.
Before #87193, the canonicalizer in arith crashed when attempting signed
extension on an i0 value. To hopefully avoid it happening again, this PR
introduces tests for canonicalization of arith operations with i0
values, focusing on operations related to bit width or signedness.
Adding `min` that was already implemented but not exposed.
Adding a few additional unary ops:
* Reciprocal as `arith.div(1,arg)`
* Round as `math.round(arg)`
* Sqrt as `math.sqrt(arg)`
* Rsqrt as `math.rsqrt(arg)`
* Square as `math.powf(arg, 2)`
* TanH as `math.tanh(arg)`
All with the agreed semantics at the round table: no implicit
broadcast/type cast.
This patch removes the `offloadingHandler` option from the
`ModuleToBinary` pass. The option is removed as it cannot be parsed from
textual form.
This fixes issue #90344.
There is the `MulSIExtendedRHSOne` canonicalization for
arith.mulsi_extended that is defined as follows: `mulsi_extended(x, 1)
-> [x, extsi(cmpi slt, x, 0)]`. In the implementation of this, there is
a `IsScalarOrSplatOne` constraint for the second argument. However, this
constraint does not correctly handle situation when multiplying i1
values. Therefore, an additional constraint has been added which checks
the second argument for strict positivity.
fix#88732
This PR adds support for `memref.collapse_shape` to sub-byte type emulation. The `memref.collapse_shape` becomes a no-opt given that we are flattening the memref as part of the emulation (i.e., we are collapsing all the dimensions).
In file included from /llvm-project/mlir/lib/Target/LLVMIR/Dialect/VCIX/VCIXToLLVMIRTranslation.cpp:70:
/llvm-project/build-Release/tools/mlir/include/mlir/Dialect/LLVMIR/VCIXConversions.inc:8:48: error: 'cast' is deprecated: Use mlir::cast<U>() instead [-Werror,-Wdeprecated-declarations]
VectorType vt = op.getResult().getType().cast<VectorType>();
^
/llvm-project/mlir/include/mlir/IR/Types.h:345:9: note: 'cast' has been explicitly marked deprecated here
U Type::cast() const {
^
In file included from /llvm-project/mlir/lib/Target/LLVMIR/Dialect/VCIX/VCIXToLLVMIRTranslation.cpp:70:
/llvm-project/build-Release/tools/mlir/include/mlir/Dialect/LLVMIR/VCIXConversions.inc:8:48: error: 'cast<mlir::VectorType>' is deprecated: Use mlir::cast<U>() instead [-Werror,-Wdeprecated-declarations]
VectorType vt = op.getResult().getType().cast<VectorType>();
^
/llvm-project/mlir/include/mlir/IR/Types.h:112:5: note: 'cast<mlir::VectorType>' has been explicitly marked deprecated here
[[deprecated("Use mlir::cast<U>() instead")]]
^
In file included from /llvm-project/mlir/lib/Target/LLVMIR/Dialect/VCIX/VCIXToLLVMIRTranslation.cpp:70:
/llvm-project/build-Release/tools/mlir/include/mlir/Dialect/LLVMIR/VCIXConversions.inc:32:48: error: 'cast' is deprecated: Use mlir::cast<U>() instead [-Werror,-Wdeprecated-declarations]
VectorType vt = op.getResult().getType().cast<VectorType>();
^
/llvm-project/mlir/include/mlir/IR/Types.h:345:9: note: 'cast' has been explicitly marked deprecated here
U Type::cast() const {
^
In file included from /llvm-project/mlir/lib/Target/LLVMIR/Dialect/VCIX/VCIXToLLVMIRTranslation.cpp:70:
/llvm-project/build-Release/tools/mlir/include/mlir/Dialect/LLVMIR/VCIXConversions.inc:32:48: error: 'cast<mlir::VectorType>' is deprecated: Use mlir::cast<U>() instead [-Werror,-Wdeprecated-declarations]
VectorType vt = op.getResult().getType().cast<VectorType>();
^
/llvm-project/mlir/include/mlir/IR/Types.h:112:5: note: 'cast<mlir::VectorType>' has been explicitly marked deprecated here
[[deprecated("Use mlir::cast<U>() instead")]]
^
4 errors generated.
This PR adds a new pattern to the set of patterns used to resolve the offset, sizes and
stride of a memref. Similar to `ExtractStridedMetadataOpSubviewFolder`, the new
pattern resolves strided_metadata(collapse_shape) directly, without introduce a
reshape_cast op.
Addresses issue #88716.
Some function parameter names in the affected header files did not match
the parameter names in the definitions, or were listed in a different
order.
---------
Signed-off-by: Troy-Butler <squintik@outlook.com>
This reverts commit b77416ea15. While that seemed like an
appropriate rename at the time, further review on pull request #90078
makes it unnecessary. This revert changes the name back.
Canonicalization defaulted to replacement when the input dims were from
unknown source. This is obviously incorrect. Tweaked and included test
to prevent future issue.
Rename `OutgoingNotification` to `OutgoingMessage`, since the same
callback function type will be used in a future commit to represent
outgoing requests, in addition to outgoing notifications.
