Digit-Life has a news item news item which points out that nVidia will release an nForce chipset without the GF2MX integrated video.
While this will certainly help reduce the pretty outrageous cost of this, it still faces two problems.
The board still bears an extremely high price, and the only reasonable justification for that price is the Dolby Digital audio. I don’t know if that alone is enough to float enough people’s boats.
More importantly, the two main performance features the nForce has over the KT266A either aren’t quite needed yet, or don’t seem to do much good.
HyperTransport is all well and good, but it doesn’t really bring any improvement to the table right now over more interim solutions.
Dual DDR may get memory maniacs all hot and bothered, but outside of a bit higher score, it’s not doing too much, either.
Throwing more bandwidth at a CPU than it can handle only has marginal impact on performance. The performance improvement of dual over single-bank DDR in the nForce very much resembles that from PIII DDR boards: next to nothing.
Until AMD comes up with a processor that can look at dual-bank DDR and say, “Please sir, I want some more,” this is an answer looking for a problem.
Anything I’ve heard about Thoroughbred seems to be saying this is just a die-shrink with no enhancements, so this probably isn’t going to help nVidia. Barton might be a different story, but I’ve seen nothing to indicate that this will change for it, yet.
On the other hand, the stray word here and there about Claw/Sledgehammer seems to indicate that AMD will definitely do something to increase its data-chewing ability. Not necessary dual-DDR, but something.
Perhaps nVidia is practicing with the dual-bank technology now to have it ready for when it’s really needed.
In the meantime, I don’t see any good reason for anyone outside of those who want Dolby Digital to prefer this generation of boards, integrated core or not.
For system builders, I think the second generation nForce chipset is the one to await. A GF3 integrated core will leave buyers less behind the video curve than the current GF2MX core, and increase the useful life of the board. It should provide a pretty good integrated solution for those Joe SixPack with kids getting to the trigger-happy stage.
I can see spending almost $200 for that. Not until then, though.