The Javadoc for the OpenGL ES API does a good job explaining why you can't use wrapped arrays for everything: > http://download.oracle.com/javame/config/cldc/opt-pkgs/api/jb/jsr239/index.html > Because these functions do not consume the data located at the given pointer > immediately, but only at some unspecified later time, it is not possible to > use an array since the Java virtual machine may relocate the array within > system memory.
But if you really want hardware details, here's a nice document for the type of graphics in the Droid: http://www.imgtec.com/factsheets/SDK/POWERVR%20SGX.OpenGL%20ES%202.0%20Application%20Development%20Recommendations.1.1f.External.pdf There's lots of stuff for older chips like the MSM7200 in the G1 out there as well, although you may have to look in a BREW oriented document or two to find it. On Sep 1, 7:31 pm, Samsyn <[email protected]> wrote: > So, I've been a good boy (I think) and used ByteBuffers (and > ShortBuffers and IntBuffers etc) for all the vertex lists and textures > I feed to GLES. > > But can someone point me to the basic architecture of the 'typical' > mobile GPU (or, I guess, the OpenVR Droid GPU at least). For example, > does a mobile GPU have its own texture/buffer memory, or does it > always share 'system memory' ? > > If it's going to copy the data I give it into 'gpu ram' anyway, then > have I just added another buffer in the bucket brigade of memory when > I use a ByteBuffer? > > GLES seems perfectly happy using a direct java array for the data, but > I imagine that does nasty things like locking memory when the GPU > touches it, when maybe memory outside the garbage collection heap > doesn't have that problem (and just gets leaked when I mis-manage > it :-) > > I can't say that I see any performance difference between using > ByteBuffers or just direct java arrays. But most of my meshes (and > all of my textures) are created just once. But some meshes (for > example, particles) get regenerated every frame. And for that, I do > the math in a java array and then allocateDirect a suitable > <whatever>Buffer before passing it to GLES. That smells really awful > to me and I guess I am just wondering how big a sin it would be to not > use <whatever>Buffers for these dynamic meshes? > > I guess I just feel like I am flying blind without some GPU > documentation to look at. -- You received this message because you are subscribed to the Google Groups "Android Developers" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/android-developers?hl=en
