llvm-project.git/mlir/test/python, branch main Unnamed repository; edit this file 'description' to name the repository. [mlir][py][c] Enable setting block arg locations. (#169033) 2025-11-21T13:31:46+00:00 Jacques Pienaar jpienaar@google.com 2025-11-21T13:31:46+00:00 5ab49edde282814f41b90431194afaff694deba7 This enables changing the location of a block argument. Follows the approach for updating type of block arg. 
This enables changing the location of a block argument. Follows the
approach for updating type of block arg.
 [MLIR][XeGPU][TransformOps] Add slice_dims argument to set_op_layout_attr and set_desc_layout (#168929) 2025-11-21T08:08:12+00:00 Tuomas Kärnä tuomas.karna@intel.com 2025-11-21T08:08:12+00:00 e9fc393a9e431d1a0aebc3fe448f3cf1668fbb34 `set_op_layout_attr` and `set_desc_layout` transform ops wrap `xegpu.layout` in an `xegpu.slice` attribute if `slice_dims` argument is set. 
`set_op_layout_attr` and `set_desc_layout` transform ops wrap
`xegpu.layout` in an `xegpu.slice` attribute if `slice_dims` argument is
set.
 [MLIR][Python] Add arg_attrs and res_attrs to gpu func (#168475) 2025-11-18T15:55:11+00:00 Asher Mancinelli ashermancinelli@gmail.com 2025-11-18T15:55:11+00:00 47d9d735a7aef937256536af490876879c4b4731 I missed these attributes when I added the wrapper for GPUFuncOp in fbdd98f74f0d. 
I missed these attributes when I added the wrapper for GPUFuncOp in
fbdd98f74f0d.
 [MLIR][Transform][Python] Expose applying named_sequences as a method (#168223) 2025-11-15T18:31:17+00:00 Rolf Morel rolf.morel@intel.com 2025-11-15T18:31:17+00:00 eb9d56cb558cd7412edc8488f331517f3bb3df30 Makes it so that a NamedSequenceOp can be directly applied to a Module, via a method `apply(...)`. 
Makes it so that a NamedSequenceOp can be directly applied to a Module,
via a method `apply(...)`.
 [MLIR][Python] Add tests for nvvm barrier ops (#167976) 2025-11-15T16:57:41+00:00 Asher Mancinelli ashermancinelli@gmail.com 2025-11-15T16:57:41+00:00 a4e7d150ea69753982bc8a1dcce310c455a069a6 Found this issue #167958 when adding these tests, thanks for the quick fix @clementval. 
Found this issue #167958 when adding these tests, thanks for the quick
fix @clementval.
 [mlir][ROCDL] Refactor wmma intrinsics to use attributes not operands where possible (#167041) 2025-11-14T00:50:02+00:00 Muzammiluddin Syed muzasyed@amd.com 2025-11-14T00:50:02+00:00 b1262d13e0e044b3a067e083989f7367496aefbe The current implementation of the WMMA intrinsic ops as they are defined in the ROCDL tablegen is incorrect. They represent as operands what should be attributes such as `clamp`, `opsel`, `signA/signB`. This change performs a refactoring to bring it in line with what we expect. --------- Signed-off-by: Muzammiluddin Syed <muzasyed@amd.com> 
The current implementation of the WMMA intrinsic ops as they are defined
in the ROCDL tablegen is incorrect. They represent as operands what
should be attributes such as `clamp`, `opsel`, `signA/signB`. This
change performs a refactoring to bring it in line with what we expect.

---------

Signed-off-by: Muzammiluddin Syed <muzasyed@amd.com>
 [MLIR] [Python] `ir.Value` is now generic in the type of the value it holds (#166148) 2025-11-13T13:23:40+00:00 Sergei Lebedev 185856+superbobry@users.noreply.github.com 2025-11-13T13:23:40+00:00 31536e6e9a54f54b26f97ce227d294d29b2b845f This makes it similar to `mlir::TypedValue` in the MLIR C++ API and allows users to be more specific about the values they produce or accept. Co-authored-by: Maksim Levental <maksim.levental@gmail.com> 
This makes it similar to `mlir::TypedValue` in the MLIR C++ API and
allows users to be more specific about the values they produce or
accept.

Co-authored-by: Maksim Levental <maksim.levental@gmail.com>
 [MLIR][XeGPU][TransformOps] Add convert_layout op (#167342) 2025-11-12T18:57:51+00:00 Tuomas Kärnä tuomas.karna@intel.com 2025-11-12T18:57:51+00:00 7f4a3a98a262f019067c94f5f47e40c5ec1ed380 Adds `transform.xegpu.convert_layout` transform op that inserts an `xegpu.convert_layout` op for a given `Value`. 
Adds `transform.xegpu.convert_layout` transform op that inserts an
`xegpu.convert_layout` op for a given `Value`.
 [mlir][linalg][python] Add Python Bindings for Inferring Contraction Dimensions from Affine Maps (#167587) 2025-11-12T18:35:04+00:00 Bangtian Liu liubangtian@gmail.com 2025-11-12T18:35:04+00:00 a5a78d0bb4198b4280cc2523841609dca82de2eb This PR exposes `linalg::inferContractionDims(ArrayRef<AffineMap>)` to Python, allowing users to infer contraction dimensions (batch/m/n/k) directly from a list of affine maps without needing an operation. --------- Signed-off-by: Bangtian Liu <liubangtian@gmail.com> 
This PR exposes `linalg::inferContractionDims(ArrayRef<AffineMap>)` to
Python, allowing users to infer contraction dimensions (batch/m/n/k)
directly from a list of affine maps without needing an operation.

---------

Signed-off-by: Bangtian Liu <liubangtian@gmail.com>
 [MLIR][XeGPU][TransformOps] Add insert_prefetch op (#167356) 2025-11-12T10:24:23+00:00 Tuomas Kärnä tuomas.karna@intel.com 2025-11-12T10:24:23+00:00 3c52f536902b1f4096e25e0e73bc3c26355cbf40 Adds `transform.xegpu.insert_prefetch` transform op that inserts `xegpu.prefetch_nd` ops for the given `Value` in an `scf.for` loop. 
Adds `transform.xegpu.insert_prefetch` transform op that inserts
`xegpu.prefetch_nd` ops for the given `Value` in an `scf.for` loop.