[Mlir-commits] [mlir] 3be05d8 - [MLIR] Fix test on Windows

Mehdi Amini llvmlistbot at llvm.org
Thu Feb 29 15:09:23 PST 2024


Author: Mehdi Amini
Date: 2024-02-29T15:09:18-08:00
New Revision: 3be05d898ffd1e687bd67dcd5fe0c8c16ff2d4d7

URL: https://github.com/llvm/llvm-project/commit/3be05d898ffd1e687bd67dcd5fe0c8c16ff2d4d7
DIFF: https://github.com/llvm/llvm-project/commit/3be05d898ffd1e687bd67dcd5fe0c8c16ff2d4d7.diff

LOG: [MLIR] Fix test on Windows

Windows folds to 1.01563 and linux to 1.01562, let's just check the prefix here.

Added: 
    

Modified: 
    mlir/test/mlir-cpu-runner/expand-arith-ops.mlir

Removed: 
    


################################################################################
diff  --git a/mlir/test/mlir-cpu-runner/expand-arith-ops.mlir b/mlir/test/mlir-cpu-runner/expand-arith-ops.mlir
index 0bf6523c5e5d5c..2b62b8c0bb140f 100644
--- a/mlir/test/mlir-cpu-runner/expand-arith-ops.mlir
+++ b/mlir/test/mlir-cpu-runner/expand-arith-ops.mlir
@@ -24,7 +24,7 @@ func.func @main() {
   // Note: this is a tie (low 16 bits are 0x8000). We expect the rounding behavior
   // to break ties "to nearest-even", which in this case means upwards,
   // since bit 16 is set.
-  // CHECK-NEXT: 1.01562
+  // CHECK-NEXT: 1.0156
   %value_1_01172_I = arith.constant 0x3f818000 : i32
   %value_1_01172_F = arith.bitcast %value_1_01172_I : i32 to f32
   call @trunc_bf16(%value_1_01172_F): (f32) -> ()


        


More information about the Mlir-commits mailing list