Uh, VLD??? Maybe $1M. We're not talking about a very complicated block. Integrating the MCP was probably more of a hassle. The memory bandwidth of the Rage Pro 128 probably already supports full HD if it could just be redirected to video decode.
General Instruments balked at their earlier attempt to push a Celeron and the RAGE 128 as a high end settop. The settop box business demands that interlaced video be dealt with solidy (not just fluffy DVD movies which are much easier to decode on a PC). So, ATI went back to the drawing board and started a design that does integrate the full decoder. This took some considerable pushing.
Thomson's SOC (no x86) decoder sells for roughly $15 and integrates NTSC out, Audio decode, system transport decoder, and video decode. In the 2M transistor range. I think quantity and fabs are the issue. DVD players, digital cable, and direct broadcast satellite drive these quantities.
Nvidia has passed DVD WHQL recently. Do your homework.
There are at least three types of SOC proponents approaching the same market from different angles: graphics controller companies trying to add CPU, CPU companies trying to add graphics, and media/communications companies adding both CPU and graphics.
I think if any, the third category will win the target market (low end media PC) because they already do the part neither the CPU or grapics companies can manage well, and they have the money (with inflated stock) to buy the right pieces they don't already have.
I'd laugh if either Intel or ATI could do a decent video decoder, video modulator, encoder, cable modem, home network modem, ADSL, firewire controller, etc.
So SOC will make a wonderfully low-cost web surf'n box. |
