RyanRio wrote: > How does listing the types of asyncmarks in a wait_async make it overly > verbose? This is what you would normally expect from such type of > source-destination pairs, you don't usually have a destination with an > "accept all" wildcard. This introduces extra steps in figuring out what the > source is, and makes the IR more ambiguous which is never a good thing. Besides this is what we already do with waits, we don't just use a wait ALL, we specifically list which counters we are waiting for.
I don't get this, we currently DO have a wait "all" (all being async ops that were part of the mark). Why is it bad to maintain that possibility https://github.com/llvm/llvm-project/pull/220442 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
