[llvm] [APFloat] Add APFloat support for E8M0 type (PR #107127)

[Thorsten Schütt via llvm-commits]

================
@@ -4626,6 +4742,11 @@ void IEEEFloat::makeInf(bool Negative) {
 }
 
 void IEEEFloat::makeZero(bool Negative) {
+  // The E8M0 type cannot represent the value zero.
+  if (semantics == &semFloat8E8M0FN) {
+    assert(false && "This floating point format does not support Zero\n");
----------------
tschuett wrote:

Firstly, assert(false) shouldn't be in the code base. To me it sounds more like a no-op.

https://github.com/llvm/llvm-project/pull/107127