New Gtx260 in September?

Rumours gotta love them... yeah and it's gonna be DX10.1 compliant, GDDR5, etc, etc.

Wouldnt that have to come with a 480bit bus?

Read somewhere that it might be called gtx270 to avoid confusion...but it is nvidia afterall

Im thinking nVidia should cal it.......G280- heheh

DX10.1 support <-- untrue Nvidia already told many times they will not support directx 10.1 only 11

yes it MAY or may not use GDDR 5 and more sp... it was aimed to be faster than HD 4870 at same price point

It will have 216 SPs from 192 where GTX280 has 240

Expreview goes on to say the new GTX 260 will retain the same 65nm GT200 GPU, clock speed, memory interface width, and memory capacity as its predecessor

http://www.techreport.com/discussions.x/15358

The problem is it will shorten the gap between GTX280 <-- the early birds may get angry for buying the costly model

Way too early for DX11. Maybe 55nm and higher clocks. Not so sure about added shaders tho. And no DX10.1

Right now the added cluster looks like a stop-gap because the 55nm refresh seems to be grievously delayed. This unfortunately does little to help the production cost issue, and really greatly devalues the GTX280 by putting any potential GTX260+ too close to it.

BTW, as for DX10.1 / GDRR5 I think some people need to research the term sarcasm first.

Adding GDDR5 to the GTX260/280 offers very little since they aren't bandwidth limited so much as shader limited, and adding DX10.1 is a complete core-redesign (not a tweak like ATi's HD2K->HD3K) and DX11, it a pipe dream for 2008 for either company.

I can see a GTX260+ coming out, but considering the cost issues, it is more an indication that the rumours of the 55nm refresh being delayed than anything great. The HD4870 trades blows with both the GTX260 and GTX280 and costs significantly less, nVidia needs something quickly to maintain performance, but reduce costs, that this is the back-to-school launch and not the 55nm would be bad, despite probably selling reasonably ok.

The Man speaks sense, He's a witch Get him!

I already rallied the mob! We got to burn someone right NOW!

Hey, TGGA. Wanted to ask you this: since current ATI cards already have a built-in tesselation engine, would it be possible for them to support DX11? I'm aware there are many other requirements in DX11 specifications, but could that engine be put to good use or would it need a complete core redesign to support DX11? Does it have to be "fully compliant"? Thanks in advance.

LOL

What is your favourite colour?

Not completely, but when implemented in DX11 games then the tesselation component should be exposed and useable. Tesselation as an optional feature is already exposed in DX10.1, and has been programmable back to DX8.1, just requiring special tweaking. What DX11 changes is that it adds the tesselator as a stage in the pipeline, whether it's filled with any instructions is a different story, because just like you can make a game completely out of vertex shader instructions and no pixel shader instructions, you could also just ignore the tesselation stage.

I don't know if it would require a complete re-design, but it depends on how easy it is to add the functions. I would think it wouldn't require a major redesign, but it would not simply as the HD2K->HD3K hop IMO. The main thing is that as a subset they would be able to implement the geometry amplification aspect of the tesselator, even without the other two component. Also you would have support for the other factors that are present in both DX10.1 and DX11, just like those that are in DX10 now and DX10 then.

that tells us Nothing^ at all. lol!