================
@@ -0,0 +1,159 @@
+// REQUIRES: aarch64-registered-target
+
+// DEFINE: %{optimize} = opt -passes=mem2reg,tailcallelim -S
----------------
banach-space wrote:In terms of priorities: 1. Preserve the original `CHECK` lines (i.e. just copy them across) 2. Minimise the set of passes that are run. So, if we can remove some passes without affecting the original `CHECK` lines, lets do it. In the past, we would actually _add_ passes to simplify the generated IR, but looks like SVE is already fairly optimised. Thanks for working on this! https://github.com/llvm/llvm-project/pull/230083 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
