NVIDIA Patent: G70/G80/RSX?

[j^aws]

Programmable graphics processor for generalized texturing

Abstract

A programmable graphics processor including an execution pipeline and a texture unit is described. The execution pipeline processes graphics data as specified by a fragment program. The fragment program may include one or more opcodes. The texture unit includes one or more sub-units which execute the opcodes to perform specific operations such as an LOD computation, generation of sample locations used to read texture map data, and address computation based on the sample locations...

...The fixed function computation units for performing texture mapping are configured in a pipeline that is dedicated to performing the texture mapping operations specified by texture map instructions. When texture map instructions are not used to process graphics data, the pipeline is idle. Likewise, when many texture map instructions are executed to perform texture mapping operations, a bottleneck may develop in the pipeline, thereby limiting performance.

Accordingly, it would be desirable to provide improved approaches to performing texture operations to better utilize one or more processing units within a graphics processor....

Programmable graphics processor for generalized texturing

Haven't had time, just skimmed it, but are they describing G7x/ RSX Texture units or something more? Maybe upcoming in a future part, G7x/RSX/G80...?

Edit: tidied url.

 

[j^aws]

Mods, please could you move this to the console tech section, thanks.

 

[ROG27]

What is this referring to?...

1. A graphics processor for processing graphics data, comprising:

a texture unit, including a load unit configured to receive address data, read data corresponding to a region from memory, and produce processed graphics data; and

a programmable execution pipeline coupled to the texture unit, the programmable execution pipeline including at least one multithreaded processing unit configured to process the graphics data and the processed graphics data, the texture unit further including a filter unit configured to receive tap data from the at least one multithreaded processing unit and to produce the processed graphics data.

What is all this speak of multithreading?

 

[Carl B]

I think they're just refering to a quad unit there, or am I missing somethig?

 

[ROG27]

Also:

2. The graphics processor of claim 1, wherein the at least one multithreaded processing unit includes a storage resource. [local storage a la Cell SPE except on the GPU?]

3. The graphics processor of claim 2, wherein the storage resource is configured to store the processed graphics data. [stores processed data relating to graphics from CELL locally]

????

 

[ROG27]

Maybe each Pixel Quad has its own local store/cache?

 

[j^aws]

Maybe each Pixel Quad has its own local store/cache?

G70 quads already have local L1 cache. I'm not sure, but they seem to be describing unified, multi-threaded shader units with coupled texture units...

 

[Shifty Geezer]

Sounds to me like G70 pipes where the texture units are tied in with the shader units, so you either texture or shade but not both, the idea being that if you're not texturing, a specialist texture-unit is silicon sitting idle. Multithreading is reminiscent of Xenos' threading to overcome stalls. I don't know how GPU's handle that already. I guess the idea is if a shader needs a texture read and the texture isn't local, rather than stall the shader pipeline waiting for the fetch, the pipe (processor. quad?) switches switches to another shader thread. That's going solely by the abstract Jaws posted and no investigation of the actual patent, mind!

 

[ROG27]

G70 quads already have local L1 cache. I'm not sure, but they seem to be describing unified, multi-threaded shader units with coupled texture units...

There also seems to be cache within the texture unit, as well...

http://patimg1.uspto.gov/.piw?docid=US006987517&PageNum=5&IDKey=1BDBDB8BD74E&HomeUrl=http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2%2526Sect2=HITOFF%2526p=1%2526u=/netahtml/search-bool.html%2526r=1%2526f=G%2526l=50%2526co1=AND%2526d=ptxt%2526s1=nvidia%2526OS=nvidia%2526RS=nvidia

 

[cho]

NV3X-style Full FP32 shader, NV4X/G7X-style ALU1 .

 

[!eVo!-X Ant UK]

According loads of website's, G80 is unified with shader model 4.0

Here's just one link : http://www.nforcershq.com/article3503.html

EDIT : WTF were did that post that this answear was ment for go???

 

[Arun]

Ant UK, you're linking to an article linking to an article linking to an interview and speculating based on it. VR-Zone has no clue whatsoever WTF they were talking about in that article. R580 vs G80? Please... I'd suggest you look at my track record, and VR-Zone's, then consider who you'd rather believe.

Uttar

 

[mckmas8808]

Am I crazy or is the timing of this patent a little weird?

 

[Carl B]

It *is* weird. Man I hate this, my name's on a patent that's been at the PTO for years and seems like it'll be there for a couple of years more. Meanwhile these big companies file stuff, pay some extra cash, and it just sails on through...

It's the definition of frustration.

 

[RavenFox]

Nay I think its on time.

 

[mckmas8808]

Nay I think its on time.

On time for what?

 

[ROG27]

Sounds to me like G70 pipes where the texture units are tied in with the shader units, so you either texture or shade but not both, the idea being that if you're not texturing, a specialist texture-unit is silicon sitting idle. Multithreading is reminiscent of Xenos' threading to overcome stalls. I don't know how GPU's handle that already. I guess the idea is if a shader needs a texture read and the texture isn't local, rather than stall the shader pipeline waiting for the fetch, the pipe (processor. quad?) switches switches to another shader thread. That's going solely by the abstract Jaws posted and no investigation of the actual patent, mind!

What you are saying makes perfect sense. I'm almost certain it has to do with multithreaded execution units and the utilization of local stores/cache to ensure the texture unit is fully utilized/running with maximum efficiency.

 

[RavenFox]

It *is* weird. Man I hate this, my name's on a patent that's been at the PTO for years and seems like it'll be there for a couple of years more. Meanwhile these big companies file stuff, pay some extra cash, and it just sails on through...

It's the definition of frustration.

Hey I know what you mean lol. I work for a patent&trademark firm in NY.

 

[RavenFox]

On time for what?

Well I think Nvidia ia about to reveal a few things quite soon and this patent might solidfy some up and comings.

 

[Carl B]

Hey I know what you mean lol. I work for a patent&trademark firm in NY.

Well maybe we should've used you guys instead!

But I live in DC so... I mean it only makes sense for us to have used a DC firm.