[Mlir-commits] [mlir] [mlir][arith][transforms] Adds Truncf f32 to f4e2m1 (PR #144157)

[Jakub Kuderski]

================
@@ -3,18 +3,29 @@
 // RUN: mlir-opt %s --convert-vector-to-llvm \
 // RUN:             --convert-func-to-llvm \
 // RUN:             --arith-expand="include-f4e2m1=true" \
-// RUN:             --convert-arith-to-llvm | \
-// RUN: mlir-runner -e entry --entry-point-result=void \
-// RUN:             --shared-libs=%mlir_c_runner_utils | \
-// RUN: FileCheck %s --match-full-lines
+// RUN:             --convert-arith-to-llvm -reconcile-unrealized-casts | \
+// RUN:   mlir-runner -e entry --entry-point-result=void \
+// RUN:               --shared-libs=%mlir_c_runner_utils | \
+// RUN:   FileCheck %s --match-full-lines
 
-func.func @check_extf(%in : f32) -> () {
-  %trunc = arith.truncf %in : f32 to f4E2M1FN
-  %res = arith.extf %trunc : f4E2M1FN to f32
+func.func @check_extf(%in : f4E2M1FN) -> () {
+  %res = arith.extf %in : f4E2M1FN to f32
   vector.print %res : f32
   return
 }
 
+// Note: x is sign bit
+// Format of F4E2M1: 
+// [sign bit]{1} [exponent bits]{2} [mantissa bit]{1}
+// | Binary | Value ( + / - )
+// | x000   | 0.0
+// | x001   | 0.5
+// | x010   | 1.0
+// | x011   | 1.5
+// | x100   | 2.0
+// | x101   | 3.0
+// | x110   | 4.0
+// | x111   | 6.0
----------------
kuhar wrote:

Can you point to the c++ file with the same table or some external website?

https://github.com/llvm/llvm-project/pull/144157