Brian Hook made a .plan update to follow-up on his earlier update (story), or as he calls it "the usual 'answer all the questions my last .plan update asked' update," talking further about support for 3D accelerators that are no longer state-of-the-art.

Okay, time for the usual "answer all the questions my last .plan update asked" update:

- "If you're willing to do this, why not do all kinds of workarounds so that you can support [insert orphaned graphics accelerator here]"

Because that's a lot of work, and we have to be very careful about how much we're willing to uglify our code and accept significant design limitations. We could add things like 4-bpp palettized textures to make the Permedia2 run even faster, but that would be a lot of work to help only Permedia2 users.

Our vertex lighting path helps _everyone_, but it especially helps out the lower end cards.

- "Are you going to support S3TC"

Yes. We are willing to support any compression mechanism that does not require precomputed data on our part. Right now we support the GL_S3_s3tc standard and it works quite well.

- "What are your new minimum system requirements"

This is still up in the air. While it should _run_ on any Pentium, that doesn't mean it'll run _well_ on any Pentium. We haven't done wide scalability testing, but with a small viewport and greatly reduced LOD on models then something like a P200MMX w/ Voodoo should be playable. No promises though until we have all our features implemented and have had time to do some optimization work on a wide range of hardware.

- "Would you recommend the vertex lighting mechanism for Voodoo?"

Yes, for maximum performance then I would recommend the vertex lighting mechanism and/or running at 512x384.

- "Is a V2 SLI going to have any benefits for Q3"

Obviously, more fill rate is better, but a lot will depend on the screen resolution you're running at and the quality of 3Dfx's OpenGL drivers when we ship. If their drivers aren't performing well, then a V2 SLI isn't going to be much help since we won't be fill rate limited. If their drivers are performing well, then yeah, the V2 SLI should kick ass. Keep in mind that a V2 SLI in our lightmapped case provides the equivalent of 360Mpixels/ second of rendering, which is REALLY REALLY fast.

I hope to have new benchmark numbers posted in a couple weeks after the IHVs get a fair chance to update their drivers.