SE: Since you re quite familiar with environment mapped bump mapping as seen on the G400 and in the Direct X 6 Specs can we expect you to have a full implementation in the Glaze 3D architecture?
Mika Toumi: Yes, as original inventors of the technology we definitely support it fully.
SE: As we ve gathered the Glaze 3D architecture was initially meant for licensing but now you ve decided to market it yourselves, what triggered this?
Mika Toumi: We have decided to make it to silicon, but not to become yet another 3dfx or nVidia as they sell the boards directly to end users also. We are looking for a strong parter(s) who have existing market to team with.
SE: You list your Glaze 3D as a quad-pixel pipeline @ 150Mhz with a fillrate of some 1.2Gtps, 4 x 150 only equals 600Mpps, does this mean that each of the pixel pipelines supports two texels per clock?
Mika Toumi: Exactly, every pixel has two textures for free.
SE: In your SIGGRAPH report you mention DXTC texture compression support, what kind of compression ratio/performance does this method have on the Glaze 3D, is it comparable to the S3 Savage 4 architecture that Microsoft has licensed for DirectX?
Mika Toumi: It is the S3 implementation that they licensed to Microsoft. Just like Matrox has the Environment bump-mapping that we licensed to Microsoft.
SE: You mention full-scene order independent Anti-aliasing, how is this produced, via an accumulation buffer or downsampling from rendering at increased resolution internally?
Mika Tuomi: We have accumulation buffer, downsampling and also our proprietary hardware algorithm to do full scene anti-aliasing.
SE: Will there realistically be a PCI version of the Glaze 3D? If so how would it compare to the AGP version in terms of performance?
Mika Toumi: Of course we decide that based on the market need. And the performance depends on the application. If the application is using direct AGP texturing from system memory, it won't be as fast with PCI version.
|