A DESCRIPTION OF THE PROBLEM :
Currently array zeroing elimination optimization only happens when a large array allocation is followed by System.arraycopy().

It would be great to have the same optimization for allocations followed by Unsafe.copyMemory() as well. This is a massive performance improvement for operations like array allocation followed by direct ByteBuffer's bulk get. This can also discourage projects hacking into jdk.internal.misc.Unsafe.allocateUninitializedArray().