[Mlir-commits] [mlir] [mlir][linalg] Type conversion of operands in new elementwise-op. (PR #131542)

[Andrzej Warzyński]

================
@@ -88,3 +88,41 @@ func.func @redundant_maps(%A: tensor<1x2x3x4x5xi32>, %B: tensor<1x2x3x4x5xi32>,
       outs(%C: tensor<1x2x3x4x5xi32>) -> tensor<1x2x3x4x5xi32>
   return %r : tensor<1x2x3x4x5xi32>
 }
+
+// -----
+
+// CHECK: @convert_f16_to_f32(%[[A:.+]]: tensor<16x8xf16>, %[[B:.+]]: tensor<16x8xf32>,
+// CHECK-SAME:                    %[[C:.+]]: tensor<16x8xf32>) ->  tensor<16x8xf32> {
+// CHECK: {{.*}} = linalg.elementwise
+// CHECK-SAME:       kind=#linalg.elementwise_kind<div>
+// CHECK-SAME:       ins(%[[A]], %[[B]] : tensor<16x8xf16>, tensor<16x8xf32>)
+// CHECK-SAME:       outs(%[[C]] : tensor<16x8xf32>) -> tensor<16x8xf32>
+//
+func.func @convert_f16_to_f32(%A: tensor<16x8xf16>, %B: tensor<16x8xf32>,
----------------
banach-space wrote:

Note, you are not following the naming convention documented at the top.

https://github.com/llvm/llvm-project/pull/131542