If you do what that tutorial is suggesting(draw onto a BufferedImage), you will never get hardware accelerated drawing.
Ignore that aspect of the tutorial, and draw direct to the Graphics context obtained from the BufferStrategy.
Seems like the author has mistakenly mixed the pre-BufferStrategy paradigm(Java 1.2-1.3), of using a BufferedImage as a back buffer, with the post-BufferStrategy paradigm (1.4+)
thats what i thought (about not having hardware accelaration)
though i am not so sure.... its the same as drawing my stuff to the graphics content of the BufferStrategy either straight up, or through an image
both manipulate the graphics context of the BufferStrategy, and i am not so sure that the drawing on a BufferedImage takes longer that drawing on the graphics context of the BufferStrategy
the BufferedStraegy uses VolatileImages (not so sure, have to recheck) for bufferes, so when you acquire the graphics context of a buffer which is currently not showing, i believe there is a chance that the speed of drawing on it is the same with the speed of drawing on a BufferedImage
the only difference (i think) is that if i use the BufferedImage, it needs one more step,,,,
which is useless (unless someone can find a reason why not)