Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: Sleeperz who wrote (27779)	1/8/1998 8:50:00 PM
From: Jim McMannis	Read Replies (2) \| Respond to of 1572659

CR, Thanks...so when is MMX2 coming vs AMD3D? Does MMX2 need software written for it to kick it into high gear or what? Does AMD3D? Or does any old 3D game benefit? Jim

To: Sleeperz who wrote (27779)	1/9/1998 12:11:00 PM
From: Ali Chen	Read Replies (3) \| Respond to of 1572659

CR <Coordinate transformation of Triangles can be currently done with the MMX.> Wrong. This is exactly the Intel's miscalculation. The accuracy of integer MMX is not enough to provide consistent tessellation under current picture quality requirements. Maybe for DOOM the MMX was ok, but not any more for a realistic 3D graphics. <AMD just wanted to have something out before Intel.> It is not just "something out". According to some game development circles, it is a set of instruction derived from serious research. The big part of it is so-called saturated arithmetics. <AMD-3D probably can do more powerful matrix mathematics,> It is not "probably", the K6-3D instructions were spacifically DESIGNED to provide effective recalculation of vertex coordinates in 3 dimensions. By definition it is achieved via matrix arithmetics. What makes you think that the only Intel is capable to invent something reasonable? For your info, is is commonly known that all Intel's own technical "innovations" were awkward kludges starting from 8086. Their own "advanced" project 8800 (iAPX432) appeared to be a total failure while the 8086 was a hasty workaround. The only goodies (advanced cache control, split-transaction bus on P-Pro/P-II, conditional MOV instructions, OOO execution and register renaming tricks) all were stolen ("copied") from other companies like DEC. I would think that the Merced project would definitely fail unless it would not be backed up by HP system expertise. People say: "money talks, bullshEt walks". No matter how big the Intel's money are, their shEt remains and smells like shEt. The MMX is an example.