[llvm] [AArch64][SVE] Prefer SVE over scalarization for masked load/store of bf16 vectors (PR #210986)

[Paul Walker via llvm-commits]

================
@@ -297,7 +297,7 @@ class AArch64TTIImpl final : public BasicTTIImplBase<AArch64TTIImpl> {
     if (Ty->isPointerTy())
       return true;
 
-    if (Ty->isBFloatTy() && ST->hasBF16())
+    if (Ty->isBFloatTy())
----------------
paulwalker-arm wrote:

This was only split out because the subtarget test. With that no longer necessary, please merge this with the other floating-point types below. 

https://github.com/llvm/llvm-project/pull/210986