I've noticed that in lwjgl, the functions which would require an array in opengl require bytebuffers or floatbuffers in lwjgl. is this for speed?

i need to be able to pass an array to a function i've written, and return one.

what would be faster, using a jbytearray, or converting my array to a buffer, sending it, having the method return a buffer, and converting it to an array?

Half if not all of the reason for Direct Byte buffers is for faster marshaling of data through JNI. It really is the correct way to do things with C esp if they are passing data back. Remember a java array does not look like a C array. But a C array looks like a Direct Byte buffer. More or less.