[Mlir-commits] [mlir] [MLIR][Linalg] Add aggregate ops decomposition pass and softmax decom… (PR #97582)

[Renato Golin]

================
@@ -210,32 +212,62 @@ func.func @softmax(%arg0: tensor<2x16x32xf32>, %dst: tensor<2x16x32xf32>) -> ten
 // CHECK-LABEL:      func.func @softmax(
 // CHECK-SAME:           %[[ARG0:[a-zA-Z0-9_]+]]: tensor<2x16x32xf32>, %[[DST:[a-zA-Z0-9_]+]]: tensor<2x16x32xf32>) -> tensor<2x16x32xf32> {
 // CHECK-DAG:        %[[D1:.+]] = tensor.empty() : tensor<2x16xf32>
-// CHECK-DAG:        %[[CST:.+]] = arith.constant -3.40282347E+38 : f32
-// CHECK:        %[[D2:.+]] = linalg.fill ins(%[[CST]] : f32) outs(%[[D1]] : tensor<2x16xf32>) -> tensor<2x16xf32>
+// CHECK-DAG:        %[[CST:.+]] = arith.constant 1.401300e-45 : f32
----------------
rengolin wrote:

Constant seems to have changed from largest negative to smallest positive. This may be the source of the test error?

https://github.com/llvm/llvm-project/pull/97582