Why?
I seem to recall reading that 3dfx is abandoning glide for opengl, and
that they expect that once the implementation is properly tuned that it
will be faster than the current glide stuff.
You seem to want to move away from opengl in the direction requiring
people to hard code support for individual boards into end user software.
The *only* reason I can think why this might be useful is where the system
is being pushed to the limit, and ever last fractional percent efficiency
is deemed worthwhile. Given that, you don't want to be sharing the board,
you want to own it. And, even there, you're going to have to include
support for a variety of targets -- so mesa should be one of your targets.
Or did I miss some important concept?
> Having userland manage hardware access will destroy this beauty. As I
> pointed out userland locking gives you the best performace at the cost
> of the system being unstable and having possible security holes. Its
> is just plain dangerous for low end hardware. You are in the long run
> end up with more complex tricks that will hurt performace to deal with
> these problems.
That's why you only give direct access to the hardware to a (trusted)
server, and you implement opengl as a library which talks to that server.
Doesn't mean that the server has to be X. If you've a separate monitor
hooked up to your 3d card it even makes a certain amount of sense to do
something other than X. But only a certain amount -- it's not clear to
me that anyone is interested enough in this to make it worth deploying
a stable system based on the concept.
But, the interface defined by opengl has a lot more long term potential
than the random individual interfaces defined by all the various cards.
-- Raul- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/