Commit fdb6c0e
[X86] Add test coverage for llvm#164853 (llvm#165245)
Show examples of where truncated loads of non-constant, but known
aligned, shift amounts can be folded into address math to avoid
loads/spills of large/illegal scalar integers1 parent 978f2e5 commit fdb6c0e
1 file changed
+1672
-0
lines changed
0 commit comments