[llvm] [AMDGPU] Improve the description of asyncmark semantics (PR #202579)

[Fabian Ritter via llvm-commits]

================
@@ -10,21 +10,18 @@
 Introduction
 ============
 
-Asynchronous operations are memory transfers (usually between the global memory
-and LDS) that are completed independently at an unspecified scope. A thread that
-requests one or more asynchronous transfers can use *asyncmarks* to track
-their completion. The thread waits for each asyncmark to be *completed*, which
-indicates that requests initiated in *program-order* before this asyncmark have also
-completed.
+Asynchronous operations are operations that are completed independently at an
+unspecified scope. A thread that requests one or more async operations can use
----------------
ritter-x2a wrote:

Here and line 23 talk about "requesting async operations" whereas the following sections talk about "initiating" them. I don't feel particularly strong that the terms have to be _exactly_ the same everywhere, but I wanted to point it out as a way to make the spec a tiny bit less ambiguous.

https://github.com/llvm/llvm-project/pull/202579