nVidia ends GeForce Partner Program

[Drak3]

3 minutes ago, D13H4RD2L1V3 said:

Respond with a better product?

Nvidia might not have anything better that can be pushed out that isn't out now. It's possible that the 'Scalable' Navi being able to stack the die count to rival, say, the Titan V, while costing less.

[Razor01]

22 minutes ago, Drak3 said:

Nvidia might not have anything better that can be pushed out that isn't out now. It's possible that the 'Scalable' Navi being able to stack the die count to rival, say, the Titan V, while costing less.

 

Dude seriously?  How much you want to bet Scalable is not what you think it means?

 

Ask any graphics programmer out there, go to Unity's or UE4 forums or ask here in the programming section, why current GPU's are not able to work well in multiple GPU forms, then revisit that scalable term.

 

Every damn generation of GPU's and people pull out the same crap.  Multi GPU solves all problems, please.

 

If it was only an architectural problem, it can be solved easily (and would have been solved soon after we had the first multi GPU cards (3DFx)), its not just a uarch issue.

 

here rudimentary explanation

 

 

 

now after this, tell me how the ROP's of current uarchs are going to communicate with the control silicon of another GPU even with an interposer instead of the bus in the middle and hide that latency?  We are talking about going over to L3 Cache then transporting that data over to other L3 cache?  Hmm wait can't even get that far because currently API's don't allow this

 

Since explicit and implicit mGPU in currently the latest API's are the same as SLi and Xfire (sfr and afr), problem not solved.

[leadeater]

30 minutes ago, D13H4RD2L1V3 said:

Which makes this more odd. I know NVIDIA needs to make money, but aren’t they making a big chunk off AI and self-driving

Not really, desktop GPU market still makes the largest share by a long way for them.

[leadeater]

39 minutes ago, Razor01 said:

Since explicit and implicit mGPU in currently the latest API's are the same as SLi and Xfire (sfr and afr), problem not solved.

SLI and Crossfire don't actually have SFR, that's one thing that is new to the current set of APIs. It wasn't a thing before due to how complex and bug riddled it is if only implemented at the driver level, API cooperation is required so post processing effects can be correctly applied to both sides of the image uniformly. Would be a rather terrible gaming experience if half the screen looked different to the other .

 

Edit:

Example

[Razor01]

33 minutes ago, leadeater said:

SLI and Crossfire don't actually have SFR, that's one thing that is new to the current set of APIs. It wasn't a thing before due to how complex and bug riddled it is if only implemented at the driver level, API cooperation is required so post processing effects can be correctly applied to both sides of the image uniformly. Would be a rather terrible gaming experience if half the screen looked different to the other .

Sli as with Xfire is the linked node mode in DX12   its the implicit path.

 

You can write to each node on per frame basis (each node is one GPU) which is done automatically via drivers, or explicitly which the programmer must be very aware of what the driver is doing.

 

There is implementation differences (nuances) I'm sure but we don't know that since the drivers took care of that before (afr)

[leadeater]

58 minutes ago, Razor01 said:

That is how the law enforced, by the letter of the law, then its interpreted by the courts to make judgement.

That hasn't always been the case and it's only since late 20th century that the specific and extremely calculated letter of the law doctrine has been taken up. Intent of law had a far bigger role than it does now previously, and it's still a well argued debate that isn't going to end any time soon.

 

Personally I'm on the intent/spirit of the law side myself. It's not an all or nothing, one or the other type of thing either but currently intent of law is tipped too far out of the picture.

 

Quote

The letter of the law versus the spirit of the law is an idiomatic antithesis. When one obeys the letter of the law but not the spirit, one is obeying the literal interpretation of the words (the "letter") of the law, but not necessarily the intent of those who wrote the law. Conversely, when one obeys the spirit of the law but not the letter, one is doing what the authors of the law intended, though not necessarily adhering to the literal wording.

 

"Law" originally referred to legislative statute, but in the idiom may refer to any kind of rule. Intentionally following the letter of the law but not the spirit may be accomplished through exploiting technicalities, loopholes, and ambiguous language.

 

Quote

Interpretations of the U.S. Constitution have historically divided on the "Letter versus Spirit" debate. For example, at the founding, the Federalist Party argued for a looser interpretation of the Constitution, granting Congress broad powers in keeping with the spirit of the broader purpose of some founders (notably including the Federalist founders' purposes). The Federalists would have represented the "spirit" aspect. In contrast, the Democratic-Republicans, who favored a limited federal government, argued for the strict interpretation of the Constitution, arguing that the federal government was granted only those powers enumerated in the Constitution, and nothing not explicitly stated; they represented the "letter" interpretation.

 

Modern Constitutional interpretation also divides on these lines. Currently, Living Constitution scholars advocate a "spirit"-esque interpretative strategy, although one grounded in a spirit that reflects broad powers. Originalist or Textualist scholars advocate a more "letter"-based approach, arguing that the Amendment process of the Constitution necessarily forecloses broader interpretations that can be accomplished by passing an amendment.

https://en.wikipedia.org/wiki/Letter_and_spirit_of_the_law

 

Quote

The Spirit of the Laws (French: De l'esprit des lois, originally spelled De l'esprit des loix; also sometimes translated The Spirit of Laws[1]) is a treatise on political theory, as well as a pioneering work in comparative law, published in 1748 by Charles de Secondat, Baron de Montesquieu.[2] Originally published anonymously, partly because Montesquieu's works were subject to censorship, its influence outside France was aided by its rapid translation into other languages. In 1750 Thomas Nugent published the first English translation.[3] In 1751 the Roman Catholic Church added De l'esprit des lois to its Index Librorum Prohibitorum ("List of Prohibited Books"). Yet Montesquieu's treatise had an enormous influence on the work of many others, most notably: Catherine the Great, who produced Nakaz (Instruction); the Founding Fathers of the United States Constitution; and Alexis de Tocqueville, who applied Montesquieu's methods to a study of American society, in Democracy in America. Macaulay offers us a hint of Montesquieu's importance when he writes in his 1827 essay entitled "Machiavelli" that "Montesquieu enjoys, perhaps, a wider celebrity than any political writer of modern Europe."

 

Montesquieu spent around twenty one years researching and writing De l'esprit des lois, covering a huge range of topics including law, social life and the study of anthropology, and providing more than 3,000 commendations.[4] In this treatise Montesquieu argued that political institutions needed, for their success, to reflect the social and geographical aspects of the particular community. He pleaded for a constitutional system of government with separation of powers, the preservation of legality and civil liberties, and the end of slavery.[4]

https://en.wikipedia.org/wiki/The_Spirit_of_the_Laws

 

Just to clarify I'm not saying that letter of the law wasn't larger focus in the past but it's much more taken to extremes now, which has given rise to the massive expansion of the legal profession to accommodate it.

[Razor01]

3 minutes ago, leadeater said:

That hasn't always been the case and it's only since late 20th century that the specific and extremely calculated letter of the law doctrine has been taken up. Intent of law had a far bigger role than it does now previously, and it's still a well argued debate that isn't going to end any time soon.

 

Personally I'm on the intent/spirit of the law side myself. It's not an all or nothing, one or the other type of thing either but currently intent of law is tipped too far out of the picture.

 

 

https://en.wikipedia.org/wiki/Letter_and_spirit_of_the_law

 

https://en.wikipedia.org/wiki/The_Spirit_of_the_Laws

 

Just to clarify I'm not saying that letter of the law wasn't larger focus in the past but it's much more taken to extremes now, which has given rise to the massive expansion of the legal profession to accommodate it.

There is room to wiggle around lol, of course, but to act on something that isn't there, its not easy to do

 

I agree if there is a law being broken, it should be justifiable logically and directly. We can't attach two or three things together and say now its broken because then we have no clue what the cause and effect is.  Too many ways to counter that in court.  Grey areas are where court battles are won or lost, if they can't even start with saying this is the cause, then it becomes all grey

 

 

[Razor01]

About AFR and mGPU

 

http://32ipi028l5q82yhj72224m8j.wpengine.netdna-cdn.com/wp-content/uploads/2017/03/GDC2017-Explicit-DirectX-12-Multi-GPU-Rendering.pdf

 

Slide 24.

[leadeater]

5 minutes ago, Razor01 said:

There is room to wiggle around lol, of course, but to act on something that isn't there, its not easy to do

 

I agree if there is a law being broken, it should be justifiable logical and directly. We can't attach two or three things together and say now its broken because then we have no clue what the cause and effect is.  Too many ways to counter that in court.  Grey areas are where court battles are won or lost, if they can't even start with saying this is the cause, then it becomes all grey

Plus there's the whole simpler times require simpler rules thing, taking the current legal system back in time would be akin to taking a car back in time (ignoring the petrol issue ofc).

[Kamjam21xx]

Therea definitely people too emotionally attached to specific companies in here.

[pas008]

24 minutes ago, Kamjam21xx said:

Therea definitely people too emotionally attached to specific companies in here.

had to check in

fanboys will always be there

this reminds me of the chevy vs ford truck every year

which lasts longer vs which gives performance while I have it etc etc etc

 

simple fact here is tap beer makes you tired lol

and shots help deal with their arguments

thx uber home soon 12 mins

 

on that note gluten fanboys will accept anything given to them as long as its their brand and gluten free

 

just like you never wake your woman from your drunkeness

automatic couch

go in smooth like you meant that shit with confidence

 

and for @razor01 remember sli was introduced by 3dfx which was decades ago

latency on gpus can be solved like sli which has very little bandwidth between the two

remember mcm will get lower latency but they must write the api to accept the card as whole instead of sli profile

like i said before divider chip like what the lucid hydra 200 did

 

yay for drivers in dx12 too

oh wait rotr showed us that already didnt it?>

 

[S w a t s o n]

Nvidia's petulant and false blog post only makes me feel better about grabbing an RX 580 as my vram stopgap upgrade. I just wish ampere/turing had a true competitor, I might be forced to get one after the RX580

[Razor01]

 

 

Well it comes down to cost effectiveness.  A control chip outside of the GPU's will cost money, it will also cause redundant parts in single GPU config.  This also causes drivers to become much more complex, and the need to hide the latency means we need more storage, cache.

 

Lets take the current GPU uarchs and break down how much silicon is used in for control, around 15-20%.  For ever GPU part, we will still need that much silicon and more because of the need for more cache.  If its not possible in today's uarch because of cost reasons.  How will be doable with a smaller node.  Initially new nodes now are double the cost of current nodes.  So its not cost effective yet.  But can be as the new node matures.  If the software isn't ready for this to happen though, anyone that tries to go this method will get screwed, double the cost per chip and not working that well all the time.  Added to it memory pooling issues. 

 

Also the more modules you have the more cache you will need and more complex the drivers must be to hide the latency that is introduced.

Quote

had to check in

fanboys will always be there

this reminds me of the chevy vs ford truck every year

which lasts longer vs which gives performance while I have it etc etc etc

 

simple fact here is tap beer makes you tired lol

and shots help deal with their arguments

thx uber home soon 12 mins

 

on that note gluten fanboys will accept anything given to them as long as its their brand and gluten free

 

just like you never wake your woman from your drunkeness

automatic couch

go in smooth like you meant that shit with confidence

 

and for @razor01 remember sli was introduced by 3dfx which was decades ago

latency on gpus can be solved like sli which has very little bandwidth between the two

remember mcm will get lower latency but they must write the api to accept the card as whole instead of sli profile

like i said before divider chip like what the lucid hydra 200 did

 

yay for drivers in dx12 too

oh wait rotr showed us that already didnt it?>

 

 

 

MCM does have less latency but the first multi GPU set up was  dual chip on one PCB no need to worry about the bus bandwidth there.

[Razor01]

damn it quoted wrong post.

[D13H4RD]

2 hours ago, leadeater said:

Not really, desktop GPU market still makes the largest share by a long way for them.

And they still hold a massive chunk of that.

 

I can somewhat see GPP as a way to keep a stronghold but isn’t that.......overkill?

[D13H4RD]

Quote

GPP had a simple goal – ensuring that gamers know what they are buying and can make a clear choice.

Quote

GPP was about making sure gamers who want NVIDIA tech get NVIDIA tech.

Uhhh, isn’t that why the big fat “GEFORCE GTX” label on the box exists?

 

[leadeater]

4 minutes ago, D13H4RD2L1V3 said:

Uhhh, isn’t that why the big fat “GEFORCE GTX” label on the box exists?

 

Looks like an RX 580 to me 

 

Other than the box colouring, Geforce written on the box and the Nvidia logo but I'm still sure it's an RX 580.

[Kamjam21xx]

22 minutes ago, VegetableStu said:

sorry for not reading though all 11 pages on mobile in advance

what. WHAT KIND OF TECHNO JARGON IS BEHIND MY SCREEN NAME. TELL MEEH

 

seriously what's so hard to learn? there's a plain end (Turbo air blowers), a mid end (Armor), and a high end (Strix) in terms of decking out features and maybe component quality. There's Geforce and Radeon. There's GPU tiers in itself. What is there to demistify??

 

is this about "OUR GPUs are the only gaming capable ones" or "we only want the best GPU to be better known to be able to play most games?" because going by the latter a "Strix GT1030" wouldn't cut it like even a 1050 would isn't it?! (and thus should not be advertised as a gaming card or don't deserve the branding in the broadest sense possible despite partnering with ASUS for TUF components or something) (I know there isn't a 1030 Strix TUF and I know Asus partnering with themselves is silly)

 

or is this about using a different GPU or spec in the middle of product cycle and clearly marking out a stack for it? why would you need everyone's exculsitivity for that?!

 

like what is it then?!?! RAAAAAAHH

Its Nvidia bs and lies.

[D13H4RD]

43 minutes ago, VegetableStu said:

sorry for not reading though all 11 pages on mobile in advance

what. WHAT KIND OF TECHNO JARGON IS BEHIND MY SCREEN NAME. TELL MEEH

 

seriously what's so hard to learn? there's a plain end (Turbo air blowers), a mid end (Armor), and a high end (Strix) in terms of decking out features and maybe component quality. There's Geforce and Radeon. There's GPU tiers in itself. What is there to demistify??

 

is this about "OUR GPUs are the only gaming capable ones" or "we only want the best GPU to be better known to be able to play most games?" because going by the latter a "Strix GT1030" wouldn't cut it like even a 1050 would isn't it?! (and thus should not be advertised as a gaming card or don't deserve the branding in the broadest sense possible despite partnering with ASUS for TUF components or something) (I know there isn't a 1030 Strix TUF and I know Asus partnering with themselves is silly)

 

or is this about using a different GPU or spec in the middle of product cycle and clearly marking out a stack for it? why would you need everyone's exculsitivity for that?!

 

like what is it then?!?! RAAAAAAHH

It’s typical PR talk made in an attempt to control whatever damage it may have caused to the company’s reputation.

 

IMO, they don’t even need GPP to do what they claim. The box already has “GEFORCE GTX” written plainly and it would be easy for any Tom, Dick and Harry to go to a computer store and make out which one’s GeForce and which one’s Radeon just by seeing green or red. Not to mention that NVIDIA GPUs typically outsell AMD’s offerings by a ton (sadly), so even when AMD launches a new GPU that beats the next flagship GeForce but costs less, the sheer juggernaut of the GeForce brand would probably mean that for the mainstream, the GeForce cards will still sell.

 

Look, I love my GTX 1060 and so do a lot of my friends who have cards ranging from 960s to 2-way 1080 Ti SLI rigs, but even they agreed that GPP should not have been a thing as it isn’t beneficial to the market and also a potential headache for board partners and to an extent, even NVIDIA themselves. They should do what they’ve been doing for the past few; make strong performing graphics processing units with strong performance:watt and efficiency.

[Tenelia]

19 hours ago, D13H4RD2L1V3 said:

snip

 

Well, if someone really wants something to be right, there's a ton of ways to argue for it... Frankly, the argument here is long beaten out of the zone, and I rather let history do the talking. But even then, I'm sure people will reinterpret it however they wish.

[Carclis]

I've not got time to read through a mammoth eleven pages but has anybody yet touched on the idea that the GPP very likely still exists? Given the "under the table" nature of the whole thing, I only see this blog post as a way to stop people from talking about it, since Nvidia's initial approach of burying their heads in the sand and hoping it would go away did not work. I guess they overestimated how much their fans thought of them judging by the backlash they received across the board, but I'm sure that same level of arrogance that put them in this mess will be responsible for the GPP continuing in secrecy.

[Kamjam21xx]

I stepped out for too long, had to get an idiot girls car unstuck. 

 

But uh, yah. 

[D13H4RD]

44 minutes ago, Carclis said:

I've not got time to read through a mammoth eleven pages but has anybody yet touched on the idea that the GPP very likely still exists? Given the "under the table" nature of the whole thing, I only see this blog post as a way to stop people from talking about it, since Nvidia's initial approach of burying their heads in the sand and hoping it would go away did not work. I guess they overestimated how much their fans thought of them judging by the backlash they received across the board, but I'm sure that same level of arrogance that put them in this mess will be responsible for the GPP continuing in secrecy.

That's what I fear. 

 

Though I also think that they would have accounted for a potential whistle-blower anywhere in the line. 

[mr moose]

23 minutes ago, D13H4RD2L1V3 said:

That's what I fear. 

 

Though I also think that they would have accounted for a potential whistle-blower anywhere in the line. 

Not that I wish to make specific comments about the legality or ethical nature of GPP,  but being smart enough not to get caught is not evidence or impetus  as to somethings continued existence. 

 

 

[D13H4RD]

3 minutes ago, mr moose said:

Not that I wish to make specific comments about the legality or ethical nature of GPP,  but being smart enough not to get caught is not evidence or impetus  as to somethings continued existence. 

 

 

I don't think it even should be evidence because as it stands, we don't know if it's even continuing in secret. 

 

I guess the only way to know for sure is to wait and see what comes. The damage is already done though