Krzysztof Drewniak cdd4313183

[mlir][LLVM] Let decomposeValue/composeVale pad out larger types (#183825) ...

Currently, as pointed out in the reviews for #183405, decomposeValues
and composeValues should be able to emit zexts and truncations for cases
like i48 and vector<3xi16> becoming i32s but currently that's an assert.
This commit fixes that limitation.

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

2026-02-27 13:35:19 -08:00

..

gpu-to-rocdl-barrier.mlir

[mlir][gpu] Add address space modifier to gpu.barrier (#177425)

2026-01-26 12:08:47 -08:00

gpu-to-rocdl-hip.mlir

[mlir][GPU] Generalize gpu.printf to not need gpu.module (#161266)

2025-09-29 18:21:24 -07:00

gpu-to-rocdl-invalid-dialect.mlir

…

gpu-to-rocdl-opencl.mlir

…

gpu-to-rocdl-subgroup-id.mlir

[mlir] GPUToROCDL: lower gpu.subgroup_id to the intrinsic where possible (#179422)

2026-02-04 00:53:07 +03:00

gpu-to-rocdl.mlir

[mlir][LLVM] Let decomposeValue/composeVale pad out larger types (#183825)

2026-02-27 13:35:19 -08:00

memref.mlir

…