[Mlir-commits] [mlir] [mlir][gpu] Support `minimumnumf` and `maximumnumf` in GPU dialect (PR #226903)

[llvmlistbot at llvm.org]

llvmorg-github-actions[bot] wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-backend-nvptx

Author: Artem Gindinson (AGindinson)

<details>
<summary>Changes</summary>

Follow up on #<!-- -->221658. Add mirroring GPU reduction kinds and lower through appropriate arith ops.

Assisted-by: Codex

---
Full diff: https://github.com/llvm/llvm-project/pull/226903.diff


7 Files Affected:

- (modified) mlir/include/mlir/Dialect/GPU/IR/GPUOps.td (+9-3) 
- (modified) mlir/lib/Conversion/GPUToNVVM/LowerGpuOpsToNVVMOps.cpp (+2) 
- (modified) mlir/lib/Dialect/GPU/IR/GPUDialect.cpp (+3-3) 
- (modified) mlir/lib/Dialect/GPU/Utils/Utils.cpp (+2) 
- (added) mlir/test/Dialect/GPU/all-reduce-minmaxnumf.mlir (+21) 
- (modified) mlir/test/Dialect/GPU/invalid.mlir (+32) 
- (modified) mlir/test/Dialect/GPU/subgroup-reduce-lowering.mlir (+24) 


``````````diff
The server is unavailable at this time. Please wait a few minutes before you try again.
``````````

</details>


https://github.com/llvm/llvm-project/pull/226903