[llvm-branch-commits] [mlir] [MLIR][TOSA] Test elementwise select lowering (PR #221387)

[via llvm-branch-commits]

llvmorg-github-actions[bot] wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-mlir-tosa

Author: Taufiq Rahman (Inconnu08)

<details>
<summary>Changes</summary>

Depends on #<!-- -->220916.

Adds explicit FileCheck coverage for the TOSA `reduce_min` and `reduce_max`
`nan_mode=IGNORE` lowering to `linalg.elementwise` select.

This addresses the missing TOSA-to-Linalg elementwise coverage noted in #<!-- -->221267.

Related to #<!-- -->221267.

Assisted-by: ChatGPT

---
Full diff: https://github.com/llvm/llvm-project/pull/221387.diff


1 Files Affected:

- (modified) mlir/test/Conversion/TosaToLinalg/tosa-to-linalg.mlir (+2-2) 


``````````diff
diff --git a/mlir/test/Conversion/TosaToLinalg/tosa-to-linalg.mlir b/mlir/test/Conversion/TosaToLinalg/tosa-to-linalg.mlir
index a803ee7d99153..485a53032a3b5 100644
--- a/mlir/test/Conversion/TosaToLinalg/tosa-to-linalg.mlir
+++ b/mlir/test/Conversion/TosaToLinalg/tosa-to-linalg.mlir
@@ -2438,7 +2438,7 @@ func.func @reduce_min_nan_ignore(%arg0: tensor<5x4xf32>, %arg1: tensor<5x4xf32>)
   // CHECK: tensor.empty()
   // CHECK: linalg.fill
   // CHECK: tensor.empty()
-  // CHECK: select
+  // CHECK: linalg.elementwise kind=#linalg.elementwise_kind<select>
   %5 = tosa.reduce_min %arg0 {axis = 0 : i32, nan_mode = IGNORE} : (tensor<5x4xf32>) -> tensor<1x4xf32>
   return
 }
@@ -2456,7 +2456,7 @@ func.func @reduce_max_nan_ignore(%arg0: tensor<5x4xf32>, %arg1: tensor<5x4xf32>)
   // CHECK: tensor.empty()
   // CHECK: linalg.fill
   // CHECK: tensor.empty()
-  // CHECK: select
+  // CHECK: linalg.elementwise kind=#linalg.elementwise_kind<select>
   %6 = tosa.reduce_max %arg0 {axis = 0 : i32, nan_mode = IGNORE} : (tensor<5x4xf32>) -> tensor<1x4xf32>
   return
 }

``````````

</details>


https://github.com/llvm/llvm-project/pull/221387