Listen to the parent here, I've seen 10x performance in production Java code just using flatbuffers(and paying the marshaling costs from ByteBuffer). 50x is no

Listen to the parent here, I've seen 10x performance in production Java code just using flatbuffers(and paying the marshaling costs from ByteBuffer). 50x is not unreasonable for C/C++ code that was OO and uses a data oriented approach instead.