So, do we know anything about RV670 yet?

[fellix]

Damn, again they didn't screened the PCB around the memory chips against EMI, just like the x1950Pro boards!
Few pennies saved.

 

[AnarchX]

Naming scheme strange too, 2600.

This should be "HD2600XTX" or just RV635XT...

 

[Arnold Beckenbauer]

This should be "HD2600XTX" or just RV635XT...

Higher clocked 2600/RV630 in 55 nm?

 

[seahawk]

This sounds really good for AMD/ATI. While G92 sounds like a huge disappointment.

 

[ShaidarHaran]

You know something we don't?

Nah, just covering my ass in case they hadn't all left yet.

 

[AnarchX]

Higher clocked 2600/RV630 in 55 nm?

Maybe, but I do not expect much (RV530 -> RV535...), it should be more costs-saving and added(?) D3D10.1.

 

[Vincent]

This sounds really good for AMD/ATI. While G92 sounds like a huge disappointment.

None

Nvidia could reserve full 128 sp G92 for D8E(G92x2)

 

[vertex_shader]

This should be "HD2600XTX" or just RV635XT...

I have no idea why here MSi rv670xtx called as RX2600XTX in the thread title (RX is MSi own branding scheme, they not use the reference HD).
Above the picture comments there is rv670xtx everywhere, so maybe its a mistake :smile:

I just noticed now rv670xtx runs with 600/600 clock speed on the CCC screenshot, 2d clocks?

 

[vertex_shader]

Looks like AA performance fixed :smile:

 

[vertex_shader]

Hopefully AGP rv670 based cards coming too. (/me hide :smile: )

 

[CJ]

Looks like AA performance fixed :smile:

So where exactly in those 5 pages did they state that?

 

[AnarchX]

So where exactly in those 5 pages did they state that?

In this thread nApoleon stated it (?= something about AA-loss under control...). But I am skeptical, we have only 194mm² and there should also be added D3D10.1 and UVD.
But lets see...

 

[vertex_shader]

So where exactly in those 5 pages did they state that?

First one, translation is not bug free, i not really understand after i read again the whole comment, maybe it was too early for me to say AA is fixed, sorry :smile:

 

[no-X]

In this thread nApoleon stated it (?= something about AA-loss under control...). But I am skeptical, we have only 194mm² and there should also be added D3D10.1 and UVD.
But lets see...

External memory interface is 256bit, so ring-bus is probably halved, too. UVD at 55nm takes about 3,3mm2... Which D3D10.1 feature would cost additional transistors to implement? I think that only minor tweaks will be needed.

 

[ShaidarHaran]

So where exactly in those 5 pages did they state that?

I think it was this line:

Not to look at the top Dingdingding

or perhaps it was this:

The flowers will soon be grateful to the ah!

 

[Sound_Card]

http://www.chiphell.com/attachments/20070921_c3a42236cfe075ae95e8p7ibkjohFcbZ.jpg

Neat little picture.

 

[nicolasb]

The flowers will soon be grateful to the ah!

Other way round, surely? The "ah!" will soon be grateful to the flowers.

 

[vertex_shader]

Fruitzilla not so optimistic about rv670:

The 2900 GT isn't expected until mid October, which might indicate that the RV670 isn't as close as we expected, or it doesn't perform as well as AMD would've liked it to.

 

[CJ]

Fruitzilla forgot to mention that the HD2900GT will be launched at or around the 8th of October and will only be around for one month. But they will probably at another news item in a few hours (or tomorrow) where they mention that... They have to keep attracting people.

And the performance numbers I heard of RV670 are pretty decent. And that was on a FX62!

 

[seahawk]

More then decent. Rumor sees RV670 ahead of the competition.