Nvidia GT300 core: Speculation

[KimB]

GT300 should be nearly as flexible as a CPU according to rumors, might be pretty good at raytracing, physics and OpenCL besides pure rasterized rendering.

It wouldn't be useful as a graphics card if it were too close to that flexible. There's no question that it will be further along the line towards generalized computing, but I doubt graphics cards will ever go all the way.

 

[Voxilla]

It wouldn't be useful as a graphics card if it were too close to that flexible. There's no question that it will be further along the line towards generalized computing, but I doubt graphics cards will ever go all the way.

Well Larrabee already is there, though it may not be that good as no blip this year about it at Siggraph.

 

[Tchock]

I would wager that it's not 452mm^2 due to phk (reputable nVidia plant, but facts are facts) posting there.

 

[Arty]

Do you honestly believe people like Charlie never deal with AMD PR? Theo/Fuad ect all deal with Nvidia PR. ((and probably AMD PR)) to some extent.

But there are also other people in the industry who have information on things like this beyond just NV PR, 3rd party AIBS with close relationship's to Nvidia/AMD ect are all well known for doing a little of their own chest beatings at times.

I think Charlie made it quite clear that he has access to AMD and Nvidia shut him out completely after his expose on the bad bump. That is different from what Fuad claims, he's the anti-Charlie while Theo just plain is not in the loop.

Btw. Wheres this PR manual. I should have a copy!

You are too good for it Chris, I mean it.

Spoiler

And Nvidia realises it too, which is why you havent recieved it yet

 

[Davros]

Noo..... It's Teh evil...

 

[Jawed]

I think Charlie made it quite clear that he has access to AMD and Nvidia shut him out completely after his expose on the bad bump.

No, Charlie has contacts within NVidia.

Jawed

 

[Arty]

No, Charlie has contacts within NVidia.

Jawed

Well I was referring to official contacts, not the ones leaking bits.

 

[ChrisRay]

Jawed is right though. Nvidia's official PR does occasionally deal with him. They just dont invite him to events. I'm sure he gets alot of generic PR statements from Nvidia too.

 

[DegustatoR]

I think Charlie made it quite clear that he has access to AMD and Nvidia shut him out completely after his expose on the bad bump.

Riiiiight. They "shut him out completely" so now they just feed him the wrong data most of the time.

 

[Ailuros]

http://www.xbitlabs.com/news/video/...ill_Not_Catalyze_Sales_of_Graphics_Cards.html

Nvidia will show with the next-generation GPUs that the compute side is now becoming more important that the graphics side,

As long as the transistors invested for the compute side doesn't cut off too much performance from the graphics side it's not a problem for me.

 

[rpg.314]

It is not completely clear why Nvidia is trying to downplay the importance of DirectX 11, DirectCompute 11, technologies that enable next-generation software.

Come on, how dumb you have to be to not get this???

 

[Ollo]

<Huey Lewis> That's the power of PhysX!

 

[mboeller]

http://www.xbitlabs.com/news/video/...ill_Not_Catalyze_Sales_of_Graphics_Cards.html
.

So they retained the separate DP units, or?

 

[Shtal]

Here is interesting article:
http://www.semiaccurate.com/2009/09/15/nvidia-gt300-yeilds-under-2/

I guess goodnight GT300 till next year maybe.....


People who voted http://forum.beyond3d.com/showthread.php?t=49120&page=132 that GT300 will not be here this year, looks correct. ATI R8xx will walk on Nvidia for awhile - if this article is accurate.

 

[Ailuros]

So they retained the separate DP units, or?

Did it ever sound to you like there were more chances to abandon the idea? I'd think though that they might have gone for a combined solution this time (dedicated units and SP/DP capable ALUs).

 

[mboeller]

Did it ever sound to you like there were more chances to abandon the idea? I'd think though that they might have gone for a combined solution this time (dedicated units and SP/DP capable ALUs).

Maybe I should quote you from a few weeks back.

 

[Scali]

Come on, how dumb you have to be to not get this???

Well, depends on how that should be viewed.
I can see how they want to downplay the DX11-level functionality like SM5.0, tessellation and that sort of thing, because they don't currently offer this. That part is easy to get. The fact that they'd be willing to downplay it probably also reflects that their hardware isn't going to arrive that soon (much like with DX10.1).

DX11 as an API, and DirectCompute in itself would be silly to downplay, as they work on current DX10 hardware aswell. I think all major game engines will at least be ported to DX11, as it is easy to upgrade from a DX10/10.1 codebase, and it opens up some new functionality, even on DX9/10 devices.

 

[KimB]

Seems clear to me. They want to leverage PhysX and CUDA.

 

[2senile]

http://www.xbitlabs.com/news/video/...ill_Not_Catalyze_Sales_of_Graphics_Cards.html



As long as the transistors invested for the compute side doesn't cut off too much performance from the graphics side it's not a problem for me.

Maybe they want to sell you the GT300 for Cuda/PhsX/Compute & a separate 295 to draw stuff on your monitor.

 

[seahawk]

Maybe they want something the competition does not have. CUDA and PhysX are for NV what Eyefinity is for AMD.