In java using openGL (ES) we can directly allocate ByteBuffer like
ByteBuffer bf ;
bf.allocateDirect();
But we can not do that in case of FloatBuffer it is not aviliable , why is that ?
I was wondering if it is because of :
Byte is accessible in hardware level (as OpenGL works just above hardware unlike delvik ) and registers in hardware (hardware of GPU ) are in bytes , even floating points numbers should be stored in 4 byte register which may not be available so we cannot allocate directly , rather we should tell the buffer to allocate the memory for a block of given size and after that put the data in those blocks and treat it again as FloatBuffer.
OpenGL es is written in c. In c floats, integers etc are not fixed size like java. A float point number in java is 32 bits. Now lets examine how java uses opengl es. When you send vertices to the graphics pipeline using java you actually call c functions that do the dirty work for you. This is called ndk and you find more info here : https://developer.android.com/tools/sdk/ndk/index.html. C is translated to assembly code thus each float can have different byte size on each phone depending on cpu architecture. You use nio buffers ( more here : https://docs.oracle.com/javase/7/docs/api/java/nio/Buffer.html ) to assure that your float array sizes are BASED on your phone's cpu architecture(native order) and not on jvm fixed primitive sizes. Lastly, imagine you have a vertices array of java floats (32bit fixed size). Your cpu floats are 64 bit. If you call an opengl es function from java your program will end up crashing. Hope i helped.