RV680?

Joined
Jun 10, 2004
Messages
931
Has any more information been released on this? My HD3870's have been back ordered at Dell again, so I may just forget about it and wait a month for the 3870x2.
 
6783large4abf826403fb09ff2.png


http://www.dailytech.com/article.aspx?newsid=10033
 
It would seem,that this card will be the top product from AMD,until the Summer or early Fall of 2009,if yesterdays analyst meeting is to be beleived.
 
It would seem,that this card will be the top product from AMD,until the Summer or early Fall of 2009,if yesterdays analyst meeting is to be beleived.
No, their top will be two of them in a system. Let's hope the drivers can deliver.
 
A single card will be meant for mid/high range performance. Two in a system is ATi's idea of uber performance.
 
So instead of designing a refresh top end chip, like they have for the past 3 generations, (rv designations usually mean midrange cards, IE rv570 = x1950pro), they just stick two of their value chips together on one board, and call it R680, sounds like laziness to me. I wouldn't consider RV570 a proper refresh high end, its just a shrunken, "improved", but none the less practically identical design as R600. x1800 to x1900 was a big jump.
 
That was Manny's point. He worried both for AMD's health and Nvidia seeing this as reason not to push the envelope on a new high end single chip design.
 
So could I crossfire this with my 3870?

Yes.

And for those of you saying ATI is being lazy rolling out the R680 dual GPU solution, think about this...

They screwed the pooch with the R600, and nobody is going to let them forget that...so months later they finally get a product out that does what they wanted the R600 to do and a bit more...it's a pretty solid product, easy on power, doesn't produce an excessive amount of heat, and is reasonably priced.

It's been getting good reviews for it's price/performance ratio, and it's cheap to make compared to the R6XX or R5XX products.

But it's not a spectacular product, and they're months and months from being able to roll out a entirely new design. So, they concentrate on making lemonade with their "lemons", and put two of them on a single board. It'll still be priced hundreds lower than today's best card from Nvidia, and can on par with Nvidia's top gun, while still being cheap to produce.

Then they can spend the money on software engineers to make the drivers work well, so that the end user doesn't see an Crossfire mumbo-jumbo, he just plays games and they run fast....that same software investment can then carry over into the 3 and 4-way Crossfire setups that RD790 boards provide, making dual card Crossfire less painful than before.

And that less painful Crossfire experience might just prompt a LOT of people running C2D systems on Intel chipsets to give a pair ATI cards a try (we're seeing that already).

In the meantime, the RV6xx line is desirable, selling well, and that's putting money back into the company...that money buys them the time to roll out the R700 (which is built from the ground up to be a multi-core solution. btw.)

Trust me, Nvidia and ATI WANT to sell next-gen cards...They NEED to sell next gen cards...staying with the same generation too long just means market saturation, no new money. There's no need for the consumer to spend MORE money if the product they have is still top of the line. Doesn't matter if there's able competition or not, they have to offer something better to get previous customers to spend money.

Nvidia started this whole agressive 6-month-refresh/1-year -new-product cycle years ago and ATI and Nvidia both are starting to hit a wall, it's taking more and more engineering time to get qualitative improvements with new designs.

This is what AMD and Intel discovered years ago with CPUs...to keep Moore's Law in effect, you have to cheat...if you can't make things faster, then stick more things in and run them parallel, for the same net effect. That net effect is more performance...and more income for the manufacturer.

On the other hand, the ruthless release schedules on new GPU's have raised consumer expectations too high. One could argue that GPU's are considerably more complex than x86-based CPU's, yet we expect new GPU designs more frequently than CPU designs.

In the CPU world, new rollouts like the A64, X2, C2D and Phenom are not the rule, they're the exception...after those are rolled out you get incremental upgrades, with smaller die sizes, some modest tweaks, and faster clock speeds on a regular basis. But NOT entirely new archetecture every year. And the consumers are fine with that.

I think the consumers (and the GPU companies) are going to have to get fine with slower release schedules for GPU's, unless there's some paradigm shift in design.
 
Back
Top