As per example of snappy-java, looks like we might be able to get performance benefits, esp. on uncompress side, by using fast(er) long (64-bit) reads. Simple testing shows that we might get +20% speedup (on JDK 6, mac os), but the real trick is doing this in a way that does not require Unsafe (i.e. graceful degradation).
Naive approach didn't quite work, but there should be a way to make it all work.
As per example of snappy-java, looks like we might be able to get performance benefits, esp. on uncompress side, by using fast(er) long (64-bit) reads. Simple testing shows that we might get +20% speedup (on JDK 6, mac os), but the real trick is doing this in a way that does not require Unsafe (i.e. graceful degradation). Naive approach didn't quite work, but there should be a way to make it all work.