A Chinese web site show off RV770

[xkm1948]

Damm! How did they do that?

Well, RV770 looks pretty nasty.

http://byfiles.storage.live.com/y1pBSEinPZTog-_oJoErbOokzLBWRDYt4bPqaaaJLMO1GGlaMFEeG0GOl0ki7NVo4E_9oBe4xNC0sE


Chiphell? I wonder how did those guys in China can get their hands on new cards much earlier.

 

[amddiesel]

my guess is because Taiwan is closer to China then it is to North America

 

[TheGreatGrapeApe]

Doesn't mean much as a picture, and doesn't look much different than an HD3K.

As for why they get their hands on them earlier, you only need to think of where the cards are made to figure that out.

Brother bob who works at the AIB partner's plant in Taiwan (or mainland China) pockets a defective part out of the reject pile and sneaks it home and voila, early pics.

It's like asking why most of the spy photos for cars come from places like Detroit, Stuttgart, Wolfsburg, Hamamatsu and Hokkaido.

Because that's where they're made. Now if these photos came from Burkina Faso or Djibouti, now that would be strange/impressive.

 

[JeanLuc]

Doesn't mean much as a picture, and doesn't look much different than an HD3K.

It probably is. If people spent more time to read the front pages of Toms news and the press releases from AMD/ATI and less time on rumour sites they would know that the R700 has been postponed to 2009. Its January 08 now and I think its very unlikely that AMD already have a test card.

 

[TheGreatGrapeApe]

Yeah, but the RV770 is not the R700, just like the G92 is not the G90, so who knows.

PS, where is this phantom R700 article you speak of as I see nothing in the news section of the R700 in 2009, the last I head it had already taped out and the target dates were April to July of '08 depending on how succesful their first A0 spin is.

 

[aznstriker92]

Yeah Im pretty sure that RV770 is like the 3870 and R700 like the 2900XT. Where does it say in the news that the RV770 has been postponed?

 

[sseerrggiioo]

for me it looks like an ordinary HD 3850 without the core

the rest is the same

i think that is a big lie

 

[xkm1948]

Chinese are evil, they are communism. They have no rights to have these top graphic technology.

 

[spoonboy]

Yeah, but the RV770 is not the R700, just like the G92 is not the G90, so who knows.

PS, where is this phantom R700 article you speak of as I see nothing in the news section of the R700 in 2009, the last I head it had already taped out and the target dates were April to July of '08 depending on how succesful their first A0 spin is.

Im with the hairy one on this one. It said 2009 in the amd roadmap but that was an implied fact not implicitly stated. 2008 is as much likely as it is not right now as far as i understand.

 

[quantumsheep]

Chinese are evil, they are communism. They have no rights to have these top graphic technology.

I really hope you aren't being serious here!

 

[JeanLuc]

Yeah, but the RV770 is not the R700, just like the G92 is not the G90, so who knows.

PS, where is this phantom R700 article you speak of as I see nothing in the news section of the R700 in 2009, the last I head it had already taped out and the target dates were April to July of '08 depending on how succesful their first A0 spin is.

It was in the AMD interview on the first page I believe.

 

[TheGreatGrapeApe]

It was in the AMD interview on the first page I believe.

Then it should be pretty easy for you to link to, eh?

 

[JeanLuc]

Then it should be pretty easy for you to link to, eh?

Ok it wasn't in the interview but I did read in Toms Hardware..........sort of. http://www.tomshardware.co.uk/forum/247048-10-again-r700-delayed-till-2009-phenom-2008

 

[TheGreatGrapeApe]

The thing about that Xbit article is that the further you read into it the more it sounds like Rivas is talking about an R700 class chip on their chipset/motherboard solutions.

Anwyhoo, would be dissapointing news if it's correct.

Seems to conflict with other reports though so it's strange.

 

[badgtx1969]

People are all confused from this slide

This says nothing about release of individual products, just when the platforms will be fully unified. The R7xx GPUs are scheduled to be part of a platform in 2009, discrete R7xx products will be out before the platform is.


Not the best links:
http://techreport.com/discussions.x/13809
http://www.theinquirer.net/gb/inquirer/news/2007/12/31/r700-taped

 

[badgtx1969]

http://www.nordichardware.com/news,7188.html

RV770 - the heart of R700, fake pictures?
Written by Andreas G 10 January 2008 23:12

There have been some reports on a new chip that has appeared in recent drivers from ATI. The chip dubbed RV770 appeared without any prior notice and there was no explanation to what it could be. Speculations suggested that it would be the chip used for the next generation R700 card from ATI, maybe because of recent reports that chip was taped out more than a month ago and that samples are suppose to be up and running. The big difference between RV770 and RV670 is that the former is rumored to contain two cores instead of just one.

A user at Chinese forum Chiphell published pictures of a single-die PCB, with a layout very similar to the layout of the Radeon HD 3850, claiming it to be the first shot of RV770. There are some slight differences in the memory layout and power consumption though, and there hasn't been much talk of a single-die R700 card, but there will most likely be one, although this one is most likely not the real deal. The dual-die R700 which most reports have been about is still missing in action.

However there is little doubt that RV770 would be anything but the heart of R700. Exact specifications are still illusive, but should be quite different from what we're used to (due to the multi-chip design).

 

[TheGreatGrapeApe]

Thanks for that link, because the roadmap does make it look even more so that their are talking about the family of VPU that will become the intergrated/on-board video part of the chipset, not talking about the graphics cards.

As for the RV770 controversy, yeah it still look like the HD38xx to me, but takingthe devil's advocate position (put speculation hat on), the thing COULD be an early test board just to test the functionality of the core itself prior to making the product that will become the basis for the R700 series. So if they wanted to negate the influence of the cross-communication and similar twining issues, the best way would be to make a single core version and then the dual die / package / etc version to more easily trouble shoot issue of scaling, etc.

Just a thought, as I'm still skeptical of the pictures.

 

[pathan7]

http://www.fudzilla.com/index.php?option=com_content&task=view&id=5058&Itemid=57

 

[JeanLuc]

http://www.fudzilla.com/index.php?option=com_content&task=view&id=5058&Itemid=57

Daul core or dual dies? Theres a differance.

 

[TheGreatGrapeApe]

Yeah from early speculation it's dual core and then an R700 type product would be dual die on a single package.

That still doesn't make much sense to me, but I guess when doing the lego math it's easier and cheaper to have 2 cores, and then disable one for the entry level model and then double them up for the enthusiast model. Doing a single core - chip for the entry level and then having 4 or more of them interconnected for the Enthusiast market would be hard and make parrallelism more difficult where you lose on die communication to inter die communication or worse yet something like the R680 with inter package communication.

Still not enough information though, but from a design perspective dual-core as the base building block has some advantages.