Again, not a graphics engineer. But I know DX 10 changed a lot of fundamental things, one of which is how it reports memory being used. You know the bug where we sometimes go out to white? It's because we were checking how much memory was available on the memory card using the protocol under DX9, and under DX 10, the number would never change. So we'd keep writing more memory into it, until we'd filled up the card. Oops! I'm not sure how that's a "feature", so I've got reservations about DX10.
There are some amazingly cool things about it. For example, we had to write our own object instancing code. One of the things we found about our ships is that we originally assumed that polys or texture memory would be the big performance limiters. But it turned out that our ships were made out of many individual pieces, all of which required a single draw call. These draw calls turn out to be expensive on their own right, and we were making hundreds (50-80 is a good number). Besides re-archicteting the ships, we also wrote some instancing code so that we could write a number of identical objects with a single draw call (say, cannons). That helped tremendously, but our implementation is almost certainly not as efficient as what DX 10 has built in, because the video cards are designed with that particular implementation in mind.
So, if I had to call Friend or Foe on DX10, I'd say Foe right now. In a year or so, it'll probably be Friend.
Mac version - I'd love to do a Mac version. I think the temper of our game is ideal for a large percentage of the Mac audience. But converting over the Alchemy engine would be a huge, huge task, and that alone would probably make it prohibitively expensive. The other aspect of it is boot camp. It makes it feasible to own a Mac and play great games, but it also removes the incentive of game makers to make native Mac versions because we can always assume that the hardcore gamer crowd on that platform will be dual booting. Sorry to be a dark cloud, but that's the general view amongst developers. You have to be a company (let's call it Blizzard) that has effectively unrestrained resources and a love of the platform to get native Mac versions.
Quote:
|
Originally Posted by barbara rosa
as someone currently on a games dev course, i highly appreciate what you just said, i honestly do, so what are your thoughts on DX10, because i really have heard nothing but grief over it, especially from a number of MMO developers.
What makes OGL so much harder on big scale projects? I know ive only written small things but graphical apps none the less and found not much difficulty in porting from one OS to another. Maybe its a combination of API imporvements and also that its just a LOT better documented nowadays. Does this mean that a mac version is never going to happen?
edit: i just did what you did too
|