Glide had a distinct advantage due to the fact that MS was doing a hellish job of developing DirectX. However with the latest releases and what in store for DX8, Direct X may become the api of choice. I've been through a good bit of 3dfx cards, and though I ran Glide, I was always aggrivated with that, since it made game developers pick sides at times. Since my other machine was a TNT, then a TNT2.
Hey I remember that Chance, and it did suck when a developer made a glide only game. And I am very much looking forward to DX8. That sucker is even making 3dfx consider a ground up rebuilding of the Glide API.
i7 2600k @ 4.6 ghz by Turbo multi - MSI Z68A-GD65-G3 - 8 gig GSkill Ripjaws @ 1600 8-8-8-24 - Corsair H100 Closed loop water cooling - 2 XFX 6870 DD Crossfire - 256 gig Mushkin Enhanced Chronos Deluxe and 250 Gig WD Blue Series - PCP&C Turbocool 850 SSI
I really doubt that you would want a re-design of Glide. You see, the real advantage of Glide is that it's got very low overhead. If you re-design it, it's going to lose that advantage.
Lemme explain. DirectX was built from a crappy third party software renderer that Microsoft bought. It's not tailored for any speciffic 3D accelerated video card, and by large not even tailored for use with an 3D accelerated card at all. It's hardware agnostic. It receives the same high level calls, and then internally the driver has to convert that into operations that the card can actually do. Said conversion may not be THAT straightforward, and it may well take some time to perform.
THAT is overhead. Time spent above and beyond what the card needs to actually perform the drawing. Mind you, it doesn't have to be a huge time. But in this business microseconds already can make a difference. If you have 1 micro-second overhead per polygon, an 1000 polygon player model alone will total 1 milli-second overhead. A single 2000 polygon model, like those from Vampire, would accound for 2ms overhead. To visualize what that means, alone that could cause a drop from 100 to 80 fps, or from 50 to 45 fps.
So we're in a domain where a micro-second is already a HUGE time.
Now back to Glide. It doesn't have to be hardware agnostic. It's especially tailored for Voodoo cards. A typical Glide call does very little more than upload the parameters to the graphics chip, and return. This means very little overhead.
Also, the fact that it's a low level API, means that a good developper gets more control and can optimize the drawing process better. It's the same thing as, say, assembly. In the hands of a good coder, it can really do wonders for the performance.
Moraelin -- the proud member of the Idiots' Guild
I'm not sure if Glide is a low level API. Many say that Glide is both fast and also easy to handle for 3D-programmers. I think that means that the classes or makro-commands in Glide (which are used by the 3D-programmer) are in itself well programmed and highly efficient.
Ich bin der Geist, der stets verneint...
I am the spirit that constantly negates...