Re: [oc] Newbie asking for advice

[Rudolf Usselmann <rudi@asics.ws>]

On Thursday 06 June 2002 12:19, Andrei Alechin wrote:
> Hi
>
>  I've had ideas for designing a graphics chip/board simmering in the back
> of my head ever since the demise of 3dfx. More recently, after a
> discussion in the forums of www.3dxl.com, these ideas reared their heads
> and were now extended to designing a whole computer architecture. This
> happened after a forum discussion about the performance of the Voodoo 5
> AGP in AGP 1X and 2X mode went _WAY_ off-topic, meandering through the
> definitions of "little-endian" and "big-endian" and finally to how all
> the standards devised by Intel were nothing but cruft (the opinion of
> several of the other posters). Not to beat around the bush anymore, am I
> a complete idiot in thinking that this reinventing-the-wheel idea can be
> done?

In one way or another we are all re-inventing-the-wheel here. Most of
the cores are open/free "clean-room" implementations of what other
have one already. Withe main point here being that our implementations
are "open".

The hardest part in a rather complex project like what you are suggesting
is to come up with a mature architecture that can be used in the real world.
Looks like 3dfx couldn't do it, what makes you think you can ? Assuming
you can actually design a GPU on paper that looks promising, you will
need to find somebody who knows how to ode verilog or VHDL professionally.
All these school projects and FPGA trials are great, but to tape out and
get back real silicon takes a lot more.

Cheers,
rudi


--
To unsubscribe from cores mailing list please visit http://www.opencores.org/mailinglists.shtml