[Open-graphics] This simplifies our search... (H.264 1080i chipset)

[James Richard Tyrer]

Rogelio Serrano wrote:
> On 5/27/07, James Richard Tyrer <tyrerj at acm.org> wrote:
>> Rogelio Serrano wrote:
>> > On 5/26/07, James Richard Tyrer <tyrerj at acm.org> wrote:
>> >> Nicholas S-A wrote:
>> >> > Well, apparently this is a "world's first", so there is (most
>> >> likely) no
>> >> > point in searching for another, let alone 1080p.
>> >> >
>> >> 
>> http://www.engadget.com/2007/05/21/fujitsus-h-264-chip-encodes-decodes-in-full-hd-a-worlds-fir/ 
>>
>> >>
>> >> >
>> >> >
>> >> > We could use that, but it raises our price point a bit ;-) (we were
>> >> > aiming for about $75 in FPGAs/DSPs at one time, but it
>> >> > appears to be much more than that now)
>> >> >
>> >> > :-(
>> >>
>> >> A bit pricey, but Wal*Mart has stated their intention to sell a whole
>> >> player for $300.00:
>> >>
>> >> http://www.theinquirer.net/default.aspx?article=39091
>> >>
>> >> Therefore, there must be a cheaper chip somewhere.
>> >>
>> >
>> > can a spartan 3a dsp handle that? but thats almost $300 already.
>> >
>> I think that we should take it as a given that for a sequential
>> processor that it would be cheaper to purchase one (probably a DSP)
>> rather than making one with a FPGA.  OTOH, a massively parallel
>> operation like DCT/IDCT or special purpose functions like YUV <> RGB and
>> image scaling would better be realized with something else -- a FPGA
>> would be one option (probably the best option since discrete chips seem
>> to be on their way out).
>>
> 
> the spartan 3a dsp has a bunch of dsp's in it.

No, it doesn't have the CPU in hardware.  It does have multipliers, 
MACs, & FIR filters that would be very useful in implementing a co 
processor for a DSP to do the three functions.  Not sure what the FIRs 
could be used for.

But you said it was expensive.

-- 
JRT