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.

Technology Stocks : ATI Technologies in 1997 (T.ATY) -- Ignore unavailable to you. Want to Upgrade?

To: Carnac who wrote (3296)	5/2/1999 1:49:00 PM
From: NTT	Read Replies (2) \| Respond to of 5927

>>>The Rage Pro 128 already has MCP and IDCT. Good, but put more. A Pentium 200+ is the most expensive and unreliable parser I could imagine.<<< Look, you know the hardware game. You know as well as I that it all boils down to a matter of gates and what's the most economical way of using them. The CPU intensive stuff of MC, IDCT, YUV->RGB and even the blending for SubPict and parsing is already in the R128 hardware. On a typical stream this means full decode only takes up 20-30% CPU time on a PII/450. It would be a waste of silicon to put the parsing in the R128 when the CPU is going to be sitting idle for the most part anyhow! This is the reason a R128 costs half of what CCube's DVExplore chip costs in quantity. Of course the more silicon you throw at something, the faster it can become, but there has to be a cost-efficiency balance. The only other thing that I think might be useful to offload from the CPU might be the AC-3 and CSS. VLD, Dequant and Navigation are not particularly CPU intensive tasks on todays machines so might as well use that idle CPU! ATI has down a bang up job in this regard. Heck, some of the competition is STILL dropping frames on high end machines when you throw on a high bitstream DVD! Not only this, ATI is also leading the path by having a chip that can decode HDTV resolutions which the others can't even begin to consider! >>>Oooo, 4-tap filter. I'm impressed.<<< Well it beats the pants off of a 2 tap horizontal only that some of the competition is flaunting. 4 tap in both directions makes it CONSIDERABLY nicer. It's also a case of diminishing returns. Having 8 taps isn't necessarily going to look twice as good. Look also at the consumer players: How many have more than 4 tap bidirectional? In fact even look at C-Cube's reference design for their decoder/encoders. They like to use chips like the SAA712/BT856 which as far I know don't have more than 4 taps either. >>>I suggest that long term, the PC won't even need a hardware encoder (for digital VCR), but instead have an architecture similar to today's DVD player software/graphics-controller hybrids.<<< What do you mean long term? It's already here. It may not have the motion estimation capabilities of dedicated hardware, but ATI is already showing software where they are generating full frame rate I/P/B MPEG-2 streams on high end machines (PIII's) today! It's only a matter of time IMO before they add a motion estimation block into the core to speed up that part of the encoding process as well. >>>Not doing your homework seems to be a trait of you folks, o' wise investors.<<< You've got to be kidding me if you think that any of the competitors chips can decode DVD/HDTV better than ATI's part -- In either efficiency or quality. We're not talking about dedicated DVD decode chips which cost twice as much as a multifunctional R128 either.