[llvm] LangRef: state explicitly that floats generally behave according to IEEE-754 (PR #102140)

[Joshua Cranmer via llvm-commits]

================
@@ -3572,6 +3572,29 @@ or ``syncscope("<target-scope>")`` *synchronizes with* and participates in the
 seq\_cst total orderings of other operations that are not marked
 ``syncscope("singlethread")`` or ``syncscope("<target-scope>")``.
 
+.. _floatsem:
+
+Floating-Point Semantics
+------------------------
+
+LLVM floating-point types fall into two categories:
+
+- half, float, double, and fp128, which correspond to the binary16, binary32,
+  binary64, and binary128 formats described in the IEEE-754 specification.
----------------
jcranmer-intel wrote:

I think it would be useful to modify the Floating-point type section so that (e.g.) `half` is specifically described as "IEEE 754 binary16", in addition to the changes here.

https://github.com/llvm/llvm-project/pull/102140