RSX tap on the door

[version]

http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&p=1&u=%2Fnetahtml%2FPTO%2Fsearch-bool.html&r=4&f=G&l=50&co1=AND&d=PTXT&s1=nvidia&OS=nvidia&RS=nvidia

[Hrama]

Errr you might want to edit that one Version. Make it more clickable and all.

[sudzy]

Patent Database Search: Error

[Illmatic]

The link's still not working Version.

EDIT: It's working now, cheers.

[yoshaw]

Filed in April 2003. Meaning work on this could had progressed a lot by the time it was applied for a filing at patent office(just assuming). A breakdown of the patent by someone in the know would be very appreciated.

[cliffbo]

er... WTF... Cps... are you there? come in Cps... are you reading this? help!

[Smokey]

lol where do you find these things Version :)

[Viper]

This could be in reference to any number of nVidia graphics solutions and not necessarily the RSX.

[overclocked]

G80 related, 99% sure:)

[yoshaw]

I'm gonna agree with Viper here. Just because the patent says Multi shouldn't mean it's RSX. Besides, if it really was RSX related. There are tons of tech sites out there waiting like hungry foxes ready to pounce on this. B3D? Hexus? FiringSquad? They didn't pick the scent of this? Kinda fishy otherwise, you see.

[overclocked]

I'm gonna agree with Viper here. Just because the patent says Multi shouldn't mean it's RSX. Besides, if it really was RSX related. There are tons of tech sites out there waiting like hungry foxes ready to pounce on this. B3D? Hexus? FiringSquad? They didn't pick the scent of this? Kinda fishy otherwise, you see.

Not really from a quick look it seems like it describes the Geometryshader in DX10 that gets its geometry from the vertex shader.

[version]

http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&p=1&u=%2Fnetahtml%2FPTO%2Fsearch-bool.html&r=2&f=G&l=50&co1=AND&d=PTXT&s1=Papakipos&OS=Papakipos&RS=Papakipos

4D !

"A discrete BRDF is naturally stored in a 4D lookup table. While commercially available graphics hardware has long had 2D texture mapping capability, no hardware currently in use has the 4D texture mapping capability necessary to implement this natural representation. Even if 4D texture mapping were implemented, the tables required to hold a measured BRDF would be quite large. At four bytes per sample (single-precision floating point) and a modest sampling rate of approximately twenty samples per radian (or a parabolic parameterization of 64.times.64 samples), a BRDF requires 64 MB of memory. Since a modern graphics accelerator board has only 64 MB of total texture memory, this storage requirement is prohibitively large. Conventional texture compression algorithms can reduce the space requirement, but not by more than an order of magnitude.
"

[makeitlookreal]

I don't know if any of these patents are related to the RSX or not, but it would be very nice if at least something unique from at least one of these patents worked its way into it. Seriously, I just don't understand Sony paying NVIDIA hundreds of millions of dollars for an almost off the shelf PC part. It seems to me it would not have taken much work at all to make the few simple changes we have heard about so far. Simply taking out some ROPS, reducing the size of the bus to 128, adding some kind of adaptor for the Flexio, and a few other minor tweaks should not cost that much money.

I'm not saying that I think there is going to be a quad SLI rig inside the PS3 with a huge ammount of RAM, gigantic buses, and so fourth. But if there is *nothing* significantly new or unique in the RSX it sure seems like a waste of money on Sony's part for all that "customization" they talked about, all the money they talked about, and all the manhours spent they were bragging about.

[woundingchaney]

This could be in reference to any number of nVidia graphics solutions and not necessarily the RSX.
QFT

[makeitlookreal]

Well, with all the documents and patents flowing around I hope at least one of them is in reference to the RSX.

I wish Nerve Damage would return with any additional information he has uncovered about the RSX.

Nerve Damage... WE NEED YOU!

Darn.... double posted... sorry... don't know how to edit them both into one... anyway...

WE need Nerve Damage. He was going to bring us more info about the RSX but then he disapeared.

[woundingchaney]

I wish Nerve Damage would return with any additional information he has uncovered about the RSX.

Nerve Damage... WE NEED YOU!

Darn.... double posted... sorry... don't know how to edit them both into one... anyway...

WE need Nerve Damage. He was going to bring us more info about the RSX but then he disapeared.
What are you talking about???

You know I havent seen Nerve Damage for a while, cant remember last time I seen him come to think of it. Please tell me you didnt send him on some kind of 007 top secret infiltrate Sony and Nvidia Labs mission.:angelgrin

You

[makeitlookreal]

Many people on here know from his public posts he was made privy to some potentially interesting information about the RSX from one of his sources. He shared that on here and on other forums. You know, all that multicore stuff he talked about that some people found interesting and other people foamed at the mouth with rage about?

Well, if you know Nerve Damage he is always looking for new information, and that's all I can really say.

However, he is not only mission to infiltrate any company, corporation, or anything else. No one is on such a mission, because obviously even those that want more information about the RSX and PS3 respect Sony's rights and realize such nefarious activity is unscrupulus.

[xbdestroya]

I'm gonna agree with Viper here. Just because the patent says Multi shouldn't mean it's RSX. Besides, if it really was RSX related. There are tons of tech sites out there waiting like hungry foxes ready to pounce on this. B3D? Hexus? FiringSquad? They didn't pick the scent of this? Kinda fishy otherwise, you see.

Version is ground-zero for patents. He's also ground-zero for arbitrary associations though; so of course this may have nothing to do with RSX (and doesn't seem to).

But I'm just saying, don't sell him short on the patent front. Half the patents I've ever seen posted on B3D came from him as well, which is good to know if you're going to use that as a place you feel it should have showed up already. :smoke:

(and if you look at the date, it just cleared today)

[makeitlookreal]

I find the patents very interesting and appreciate him for sharing them with us. We don't know if they directly relate to the RSX or not, but I have a feeling Version knows more than he is telling us. Perhaps there is a very good reason why he does not say much. Just a few paragraphs of information can be ran through a computer and compared to various writings to tell who it came from. If he is in a position to know certain bits of information it's quite possible he has to stay mysterious.

Personally, if I was in a position to know something (which I am not) and wanted to share the information I would say as little as possible and find ways to drop it here and there and then quickly cover my tracks. I'm not saying this is what Version is doing, but his lack of openly communicating with us and all of these VERY timely patents he shares with us is quite interesting.

Version, regardless if your an insider trying to share important information with us or if your just an ordinary person with a knack for hunting down information thanks for sharing all of this with us. It is appreciated!

[frosty]

version, you better chill out, you're on the border of violating NDA! I almost lost my job because of what I said, remember?

[yoshaw]

Version is ground-zero for patents. He's also ground-zero for arbitrary associations though; so of course this may have nothing to do with RSX (and doesn't seem to).

But I'm just saying, don't sell him short on the patent front. Half the patents I've ever seen posted on B3D came from him as well, which is good to know if you're going to use that as a place you feel it should have showed up already. :smoke:

(and if you look at the date, it just cleared today)

Oh, I see. My apologies then. I don't see him post often but one liners here n there. After reading your post, I must give him benefit of the doubt. :smoke: *goes back to reading patents with fruitless deciphering efforts*

I'm still waiting on the breakdown of one of these patents though. Would be nice to have a layman proof patent flying around here. But I'd refrain to ask those who're under any sort of NDA(s). Everyone else, more than welcome!

[overclocked]

http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&p=1&u=%2Fnetahtml%2FPTO%2Fsearch-bool.html&r=2&f=G&l=50&co1=AND&d=PTXT&s1=Papakipos&OS=Papakipos&RS=Papakipos

4D !

"A discrete BRDF is naturally stored in a 4D lookup table. While commercially available graphics hardware has long had 2D texture mapping capability, no hardware currently in use has the 4D texture mapping capability necessary to implement this natural representation. Even if 4D texture mapping were implemented, the tables required to hold a measured BRDF would be quite large. At four bytes per sample (single-precision floating point) and a modest sampling rate of approximately twenty samples per radian (or a parabolic parameterization of 64.times.64 samples), a BRDF requires 64 MB of memory. Since a modern graphics accelerator board has only 64 MB of total texture memory, this storage requirement is prohibitively large. Conventional texture compression algorithms can reduce the space requirement, but not by more than an order of magnitude.
"

That sounds to describe the little bit where all VS ALUs performs Vector/Scalar ops (Which are all 4D) combined with a way to increase the performance of VTF and VT compared to G70 class hardware. Thats my best guess with not reading throug all. :)

You can see the patent strive after having the same same output and input also ie G7X does its texture-lockups with 2D so its seems like a fairly natural improvementfor just that function.

[makeitlookreal]

Frosty,

What NDA are you saying Version is almost breaking? Are you confirming that Version is a game developer or is employed by a company somehow connected to the RSX? The reason I ask is that so far I have not seen Version write anything about his job or occupation. With all these interesting pathents and his general behavior I would THINK he might be "in the know" somehow, but personally I have no way to know for certain. Frosty, are you trying to say that he is a person that could know these things?

[xbdestroya]

No, Version isn't a game developer with a big video game company, he's just straight-up Version. He's done homebrew on PS2 (right Version?) and is just generally tuned in to anything that might be PS-related.

But he casts a wide (very wide) net. :smoke:

Anyway I don't know what Frosty's talking about, but as for Version, he could care less about breaking NDA's, so be sure whatever he gleans goes immediately up.

That said, for those that don't understand Version's posting style, there's really no way to explain it. It's not too much use askign him to explain himself ever, because usually he's just putting abstracts or theories out there in a one-liner way, not actually dropping real-world hints as some of you have begun to believe.

[makeitlookreal]

Nerve Damage and Version seem very similiar to each other except that Nerve Damage would actually speak up and communicate more.

[HighOnSunoco]

version, you better chill out, you're on the border of violating NDA! I almost lost my job because of what I said, remember?

frosty, are you under a NDA? Because your post has me confused... :shrug:

[rog27]

Nerve Damage and Version seem very similiar to each other except that Nerve Damage would actually speak up and communicate more.

From the looks of Version's posts (no offense Version), I'd say it's safe to assume he is not a native English speaker. I'm sure he'd communicate more clearly in English if he could.

[makeitlookreal]

If he would try to communicate more (even in his own langrage) I would be willing to at least Bablefish his comments and maybe someone on here would be willing to translate them. Also, at times (rarely) he has said *more* than a few words or a phrase.

I'm not trying to be insensitive, because I don't have a problem if his english is not very good or if he wants to post in his native langrage. I just wish he would try to communicate a little more *if* doing so would not cause him to take a huge risk in losing his job or anything.

[Smokey]

:) :) version

[kaphwan]

Any further speculation regarding Version's posting style is about as useful as a one-legged man in an arsekicking contest.

It's just his way. Can't we leave this be?

And speaking of Babelfish translations, that whole patent might as well be in Chinese. It's best that I just concede it's beyond my scope and move on.

[archy121]

4D !

"A discrete BRDF is naturally stored in a 4D lookup table. While commercially available graphics hardware has long had 2D texture mapping capability, no hardware currently in use has the 4D texture mapping capability necessary to implement this natural representation. Even if 4D texture mapping were implemented, the tables required to hold a measured BRDF would be quite large. At four bytes per sample (single-precision floating point) and a modest sampling rate of approximately twenty samples per radian (or a parabolic parameterization of 64.times.64 samples), a BRDF requires 64 MB of memory. Since a modern graphics accelerator board has only 64 MB of total texture memory, this storage requirement is prohibitively large. Conventional texture compression algorithms can reduce the space requirement, but not by more than an order of magnitude.
"

Hello,

Im on the the Search for the Holy Grail Of RSX !

That quote above is MOST INTERESTING & RINGING Bells & Whistles to me ! !

But the thing is I cant find that quote on the Link page ??!

Is the link correct can anyone esle find that text on the linked page ?


Please provide me a link to exactly where that quote comes from

Thanks Version !



Archy

[frosty]

looks like Makeitlookreal has some competition, lol. Welcome!

[archy121]

G80 related, 99% sure:)


I have found the quote, just few pages on - Link works great.


Overclock why do you say with such certainty (99%) that it refers to G80 ?

Which particular part of the patent makes you say this ?


Thanks


Archy

[makeitlookreal]

I would like to introduce my friend Archy to the forums.

[archy121]

looks like Makeitlookreal has some competition, lol. Welcome!


Yes.. we knights are part of the same quest :D

But i have a feeling that i might just have found more a sign to what i have been telling him & many others on ps3forums - thanks to Version

[Viper]

:buldge:

I....err.....maybe not.

TGS guys.
*leaves thread*

[makeitlookreal]

Can anyone here tell us if the patents here are related to the G80? It would be nice to know exactly where this technology is going to be utilized. Is this some how related to NVIDIA's future G80 GPU?

[Beenie Man]

Oh shit the duo.

[xbdestroya]

Well, the PS3Forums must be a pretty ignorant place - that's what I gather from this conversation. :smoke:

[frosty]

they are, trust me.

[xbdestroya]

they are, trust me.

Noted. ;)

Anyway a lot of conversation has centered around whether this applies to RSX or not, which is a shame because the patent would be interesting to discuss in it's own right.

But basically it can be boiled down to this: Building a Better Vertex Shader

As for why it clearly is 'general' NVidia technology and not RSX specific (and not to say that RSX couldn't utilize some of the concepts), but the filing date was April of 2003 - well before NVidia and Sony formalized their relationship, which I suspect happened Fall of 2004, and was only officially announced December of 2005.

Plus if you read the patent (and I admit to only having browsed it thus far myself), it's definitely written in the context of standard GPU architectures... and just 'building a better horse' in general.


PS - By the way, Archy welcome to the forum, though I would not have believed anyone else on the net felt as passionately about the RSX as did MILR.

[venomv]

http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&p=1&u=%2Fnetahtml%2FPTO%2Fsearch-bool.html&r=2&f=G&l=50&co1=AND&d=PTXT&s1=Papakipos&OS=Papakipos&RS=Papakipos

4D !

"A discrete BRDF is naturally stored in a 4D lookup table. While commercially available graphics hardware has long had 2D texture mapping capability, no hardware currently in use has the 4D texture mapping capability necessary to implement this natural representation. Even if 4D texture mapping were implemented, the tables required to hold a measured BRDF would be quite large. At four bytes per sample (single-precision floating point) and a modest sampling rate of approximately twenty samples per radian (or a parabolic parameterization of 64.times.64 samples), a BRDF requires 64 MB of memory. Since a modern graphics accelerator board has only 64 MB of total texture memory, this storage requirement is prohibitively large. Conventional texture compression algorithms can reduce the space requirement, but not by more than an order of magnitude.
"

Umm, someone needs to explain this me, I am completely confused. Correct me if I am wrong, but the forth demension is time..........

And welcome Archy..........I think.

[makeitlookreal]

For those of you who have said (even just teasingly so) negative things about me in the past.

Two maniacs are better than one! :-)

I think that were the start of a much larger coalition!

If that patent was from 2003 it sure seems that they could have utilized some of the technology in the RSX. Could someone here give a laymens version of that patent and explain it in a way a non-tech person could understand?

Thanks so much everyone. I appreciate my E-mpire family.

[venomv]

Two maniacs are better than one!

Suprisingly enough I have to agree with that.

[archy121]

Thanks for the welcome !

Yes MILR brought the subject to my attention but i believe i have made some advancements in a new direction which i hope to share with you all soon as i can get one or two more pieces into the Jigsaw.

I need someone to decipher that patent & be able to convince that it relates to the G80 GPU -99% according to Overclock.

Please figure out if this patent is realted to G80 & to what degree.

BTW xbdestroya
Just because Nvidia filed the patent doesn't mean if Sony came along a year or two later that technology could not be used for them ?


Archy

[Beenie Man]

*cough*

EDIT: Ok, edited for the sake of cross-Playstation forum relations. :smoke:

Duly noted though Beenie Man.

[xbdestroya]

It's not that it 'relates' to one GPU or another, it's that it's technology that can be applied anywhere it would be practical to do so. For G80's part, it's decoupled texture units seem like they be right in line with using some of what seems to be discussed in this patent - so it makes sense to think G80. But again, I've only glanced over it thus far.

I'm reading it more thoroughly now, so maybe I'll have some sort of useful 'translation' in a bit. Though if Cpi shows up, obviously we'll off that task to him. ;)

[archy121]

EDIT: Edited because the post you were replying to was edited. :smoke:

[archy121]

I'm reading it more thoroughly now, so maybe I'll have some sort of useful 'translation' in a bit. Though if Cpi shows up, obviously we'll off that task to him. ;)


Top bloke !


Thanks & good night -its 3:30am for me.


Archy

[xbdestroya]

Ok, first let's put the summary of the invention out here because that really is as close to a brief synopsis as you get in a patent:



SUMMARY OF THE INVENTION

New systems and methods for the processing of graphical data are disclosed. The systems include a graphics data-processing pipeline configured to generate vertex data that can be used as input to a subsequent pass through the graphics data-processing pipeline. In various embodiments, the vertex data is generated and stored in a format suitable as an input to a geometry processor or a fragment processor. For example, in some embodiments, vertex data is saved in a texture data format. Vertex data may, therefore, be manipulated using techniques that were conventionally restricted to texture data. For example, in some embodiments vertex data is rendered in a first pass through the graphics data-processing pipeline and then used by the geometry processor or the fragment processor in a second pass.

A graphics subsystem includes a geometry processor configured to receive input vertex data and process the input vertex data received to produce vertex data output, a resampler configured to receive the vertex data output of the geometry processor and to generate a resampler output responsive to the vertex data output received, a fragment processor configured to receive the resampler output, to modify the resampler output and to generate a fragment processor output responsive to the resampler output received and a raster analyzer configured to receive the fragment processor output and to generate processed vertex data as output in a format compatible with the geometry processor.

Another graphics subsystem includes a geometry processor configured to receive and produce vertex data, a resampler configured to receive the vertex data produced from the geometry processor and to provide resampled data, a fragment processor configured to receive the resampled data from the resampler and to modify the resampled data to produce fragment data, a raster analyzer configured to receive the fragment data from the fragment processor and output vertex data in a format compatible with the geometry processor and a local memory configured to store the vertex data output of the raster analyzer.

A method of processing graphics data including receiving vertex data at a fragment processor and receiving first vertex perturbation data at the fragment processor. The vertex data received and the first vertex perturbation data is combined using the fragment processor to produce first perturbed vertex data.

Another method for processing graphics data including receiving vertex data, receiving control points representing a primitive surface contour and receiving shader program instructions. The shader program instructions are executed in a fragment processor to produce processed vertex data using the vertex data.

An apparatus for processing graphics data, including a geometry processor to receive primitives and configured to generate vertices responsive to the primitives received, a resampler to receive the vertices and configurable for a first mode and a second mode, wherein the first mode the resampler adapted to increase sample density of the primitives responsive to the vertices to provide polygon meshes and a fragment processor to receive the polygon meshes and to receive control points, the control points being associated with the polygon meshes, the fragment processor configured to produce patches responsive to the polygon meshes and the control points, the patches having surface contours.

A computing system including a host computer and a graphics subsystem. The graphics subsystem including a geometry processor configured to receive input vertex data from the host computer, to receive processed vertex data and to produce a vertex data output. The graphics subsystem also including a resampler configured to receive the vertex data output of the geometry processor and to generate a resampler output, a fragment processor configured to receive the resampler output, to modify the resampler output and to generate a fragment processor output, and a raster analyzer configured to receive the fragment processor output and to generate the processed vertex data as output in a format compatible with the geometry processor.

Another graphics subsystem including means for receiving input vertex data and processed vertex data and for producing a vertex data output, means for receiving the vertex data output and for generating a resampled output, means for receiving the resampled output, for modifying the resampled output and for generating a fragment output and means for receiving the fragment output and for generating the processed vertex data as output in a format compatible with the means for receiving input vertex data.

Now... what does it all mean, as simply as possible? Basically I read it as a patent describing NVidia's hardware solution to the DX10 geometry shading requirement. Of course it doesn't mention DX10 by name, but that is nonetheless what would form the impetus behind such a move, as geometry shading is required for DX10 compliance - and G80 is compliant...

[imported_The_One]

Well, the PS3Forums must be a pretty ignorant place - that's what I gather from this conversation. :smoke: Err... How did you gather that from this conversation :question:

they are, trust me. I feel offended :-(. Aw, well, c'est la vie.

Two maniacs are better than one! Not if they're creating a chaos :laugh:.

Yes MILR brought the subject to my attention but i believe i have made some advancements in a new direction which i hope to share with you all soon as i can get one or two more pieces into the Jigsaw.
Eh, which is what exactly? I guess you'll tell us, in due time, right ;)?