ATI DirectX 11 'Manhattan' Mobile GPUs

Not sure if this has already been posted, but it's news to me:

ATI PLANS ON UPCOMING EVERGREEN MOBILE PARTS

"ATI is reportedly planning a series of mobile GPUs shortly, nine separate mobile parts including three GPUs with three variants each. The family is called Manhattan; the chips are Broadway, Madison, and Park. Each has a high end XT, a mid range Pro, and a low end LP variant, but some of those names may change before eventually being released. Since they are part of the Evergreen series (R800), they obviously are all 40nm, DirectX 11 and should be notably faster than their mobile M9x predecessors – as reported by Semiaccurate.

Power is said to be 45-60W for the GDDR5 Broadway XT, dropping to 30-40W for the Pro, and the GDDR3 based LP takes only 29W. Madison uses GDDR5 for the 20-30W XT, either GDDR3 or 5 for the 20-25W Pro, and 15-20W for the GDDR3 based LP. Park goes down from there, 12-15W for the GDDR5 XT, 10-12W and sub-8W for the Pro and LP respectively, both of which use GDDR3. As with the 7xx parts, the memory controller will allow vanilla DDR3 in place of GDDR3 as well - informs Semiaccurate.

Broadway and Madison should be released in August and Park is expected by November. They are notebook parts, but looking at their compatible with the M9x, you might just see one or two on desktop as well." - NotebookCheck

--------------

If this is true, i'm particularly interested in the 12-15W Park GDDR5 XT GPU which could [hopefully] end up in a 13/14" notebook .
Even NVIDIA's recently announced low-end 14W DX10.1 G210M card doesn't use GDDR5. This would essentially give it a 128-bit memory bus. And it seems ATI will be the first to release a DX11 mobile part well before NVIDIA does. Just in time for Windows 7!

 

Very nice news, I can't wait to see what kind of GPUs would hit the market.

 

Holy crap. You're going to be able to do full fledged gaming on lightweight laptops at this rate XD.

 

Good numbering system. Any news on architecture yet like number of SP, TMUs, and ROPS? I hear this time around it's going to be about implementation of true multicore GPUs, with the foundation GPU being 1200 SPs at most with the high end cards seeing 2 and 4 core versions so we could see 4800 SP GPUs!

 

awesome news, thank you

 

No further information has been leaked yet on these chips. They are supposed to be 35-45% faster then the previous generation and support hybrid SLI.

It seems these GPUs are also "pin to pin compatible with the M9x designs". M9x = Mobility Radeon 4000 HD series. In other words manufacturers might not need to redesign current notebooks or wait for Calpella to be released. They could switch out older cards and replace them with these new ones as soon as they are released. Like what Dell did with their Studio XPS 16 (HD3670->HD4670).

But this has probably always been the case.

 

This means that I could pass HD4670 and jump at HD5750 or whatever its name is Well, at least if someone is good enough to release it in MXM 2.1 Type II flavor (praying really hard ).

 

Link

 

There is no point, GPUs are effectivly highly parallel machines already, there is an argument for saying GPUs are already multicore (eg each shader being a core in the way it acts).

The high end card will be just like the 4870x2.

 

I think these are the desktop cards.

Yes, GPUs are already like massive multicored systems already.

 

Except on a single die, much like Core 2 Quads which are just two Core 2 Duos with an HT interconnect. There is and was problems with running games on x2 cards due to communication issues between the two dedicated GPUs. If ATi can get multiple full R800 cores on a single die, then it could possibly be eliminated because yes, current unified shader cards are somewhat built like this, and they can be scaled to different sizes and TDPs at the cost of total capability per clock.

 

I am waiting for the DX11 cards from Nvidia.

 

Launch Q1 2010. Maybe up for sale in notebooks some time late spring to mid summer...if you're lucky.

 

It's possible they'll have a paper launch late late Q4, releasing only a few (cause that's all they'll have) GT300s, but prices will be insane. I'd say post Q1 for GT300 at Nvidia's level of "decent" prices.
http://www.semiaccurate.com/2009/07/29/miracles-happen-gt300-tapes-out/

 

That would require duplicating a lot of logic for no reason, far better to create a chip with double the shaders rather than connecting two chips.

 

agreed, a GDDR5 Park GPU packed into a 12" chassis would be amazing.

please ATI, let the low end have 160 shaders, doesn't seem unreasonable given that i estimate the new high-end desktop chips will have 1280 shaders.

 

Unless they lower the low end to 32-bit GDDR5 (comparable to 64-bit GDDR3, which is the current low end). nVidia did that with their high end GTX 250/260M. Instead of a 256-bit GDDR3 card (9800M GTS), it became a 128-bit GDDR5 card.

 

Its also just like the ATI 4860 DDR5 AND 128 bit.

 

PIKACHU!!!! sorry for the sidetrack. Good ol Digg.

 

Why thank you

 

I was actually thinking about selling my NP5796 but I might wait and upgrade the GPU if possible.

 

If they came out with an MXM 2.1 GDDR5 4870 or 58xx, not only would I be surprised but I'd be keeping my notebook for a while

 

They are confirmed for MXM 3.0b but I haven't heard anything for MXM 2.1.
I may be wrong, but is the GDDR5 based 4870 named 4870XT as opposed with the GDDR3 based 4870 in W90?

 

Awesome. ATi are really pulling it all out here .
The MSI 17inchers are compatible with mxm 3 and backward compat with 2, so hopefully we will get something to upgrade with the correct power requirements and sizing, if not custom mosd ftw.

 

not sure how this relates to my desire for 160 shaders, which incidentally is what is speculated on wikipedia.

Yay-me! let's hope its true.

 

RIP nVidia.

 

you better put up a how to if you do that or i am gonna come over there and give you indian burns and wedgies for not sharing ;D

 

How come I missed this thread? Anyways... that's great news indeed.
I just have to quote what classic77 wrote: "R.I.P nVidia"