nVidia GeForce Partner Program: Well Intention Marketing or Anti-Competitive

[mr moose]

24 minutes ago, Notional said:

Well, if you read the original article, Bennet clearly states he has talked to seven third party vendors about this. Other sites might not have, and thus won't have the same evidence as hardocp. This is what Hardocp wrote:

Now, if is entirely your prerogative not to believe any of it, but we are taking about classified documents here. Exactly what do you require?

 

I require the article to be consistent, I require the story to come from more than one source.   So far there is only one journalist making said claims and no one to back it up other than videocardz who supposedly fears a lawsuit. 

 

Also the claims don't exactly fit with the reality of the market right now.   If what they are claiming is true then they are essentially trying to insinuate the AIB partners (who are already selling all their stock) are being offered a voluntary program that won't increase their sales of nvidia products but possible reduce sales of AMD. That doesn't sound logical to me, would you accept those terms?  I wouldn't.   That's like saying sure I'll take 30% revenue hit for no gains anywhere in my business. 

 

EDIT: and I should add, that if HardOCP are under some sort of NDA then they wouldn't even be writing the article in the first place, if they aren't under NDA then there is no legal reason why they can't publish the leaked documents they claim exist.  Any other excuse for not printing them requires we construct reasons rather than the reasons being presented or even self evident.

[laminutederire]

52 minutes ago, Razor01 said:

Why did nV outsell AMD?  It started with Maxwell against the r3xx and Fury Lines.

 

r3xx was late by 6 months, Fury line a couple months after, they dropped down to 20% and less market share because of that.  Then Polaris against Pascal, status quo, they got clobbered again.  What is going to happen with next gen, yeah its going to be worse then what we saw before. 

 

How do you blame this on a consumer when AMD brings out rebranded products and late products to the market and expect consumers to just buy AMD and nV products equally?

Amd as a company suffered a lot from Intel's practices. That meant less money their r&d in both departments. Then Nvidia outsold them when Amd had superior products quite a few times ( that happened more r9 2xx series). Both facts left them without any money at the end of a cycle of innovation not rewarded. They had to choose wisely where to put their money: they put it in CPU's and it made them breathe a little.

 

(Nvidia has been sitting on its ass for quite a while anyway by the way ( need we remind you that Pascal is Maxwell on a better manufacturing node?) Not to defend AMD but they have had inferior manufacturing nodes which holds them back both on the you market and on the CPU market, but that part could change and show that Nvidia isn't that far away and that in the long run both companies have similar perf with everything else being equal, but amd may have literally lost the silicon lottery with Glo Fo instead of TSMC these last few gens. You should try to have a bit more perspective to understand the current state of the market.)

 

[Notional]

9 minutes ago, mr moose said:

I require the article to be consistent, I require the story to come from more than one source.   So far there is only one journalist making said claims and no one to back it up other than videocardz who supposedly fears a lawsuit. 

 

Also the claims don't exactly fit with the reality of the market right now.   If what they are claiming is true then they are essentially trying to insinuate the AIB partners (who are already selling all their stock) are being offered a voluntary program that won't increase their sales of nvidia products but possible reduce sales of AMD. That doesn't sound logical to me, would you accept those terms?  I wouldn't.   That's like saying sure I'll take 30% revenue hit for no gains anywhere in my business. 

I don't really see any inconsistency? But having several sources is of course ideal. Like I said, this thing isn't over yet, and I'm sure we will see more information from other sources soon. Kyle Bennet, btw, is the person who broke the news a long time ago, about an Intel CPU with AMD graphics built in. Something I thought was complete nonsense. He proved me wrong for sure.

 

I don't follow your train of thought here. NVidia's idea here is to shape the consumer mindshare to associate NVidia Geforce with gaming and subsequently AMD not with gaming (but rather crypto or irrelevant). That makes complete sense for NVidia to do. After all, this cryptocrap won't be permanent, and the markets will reestablish again. Companies operate long term, not short term after all.

And who would accept those terms? Well, companies who want access to NVidia products up front, instead of being left behind their vendor competitors. So pretty much everyone? Especially if NVidia is indeed their primary GPU income. This is where the anti competitiveness comes in. NVidia is abusing their market power to disassociate their only competitor with gaming, using the vendors as pawns.

[mr moose]

1 minute ago, Notional said:

I don't really see any inconsistency? But having several sources is of course ideal. Like I said, this thing isn't over yet, and I'm sure we will see more information from other sources soon. Kyle Bennet, btw, is the person who broke the news a long time ago, about an Intel CPU with AMD graphics built in. Something I thought was complete nonsense. He proved me wrong for sure.

 

I don't follow your train of thought here. NVidia's idea here is to shape the consumer mindshare to associate NVidia Geforce with gaming and subsequently AMD not with gaming (but rather crypto or irrelevant). That makes complete sense for NVidia to do. After all, this cryptocrap won't be permanent, and the markets will reestablish again. Companies operate long term, not short term after all.

And who would accept those terms? Well, companies who want access to NVidia products up front, instead of being left behind their vendor competitors. So pretty much everyone? Especially if NVidia is indeed their primary GPU income. This is where the anti competitiveness comes in. NVidia is abusing their market power to disassociate their only competitor with gaming, using the vendors as pawns.

 

Kyle Bennett could have landed the first man on the moon for alI care.  I am not interested in "anonymous" information.  Anyone who judges the guilt of any company based on such loose insinuations is not be fair or robust with their reasoning. 

 

So you honestly believe that as of right now with only hearsay and anonymous claims that AIB manufacturers are choosing to limit what they can sell as "gaming" solely to gain some branding which won't actually net them any more sales for the next few years?  How is that not illogical?

 

 

 

[Razor01]

14 minutes ago, laminutederire said:

Amd as a company suffered a lot from Intel's practices. That meant less money their r&d in both departments. Then Nvidia outsold them when Amd had superior products quite a few times ( that happened more r9 2xx series). Both facts left them without any money at the end of a cycle of innovation not rewarded. They had to choose wisely where to put their money: they put it in CPU's and it made them breathe a little.

 

(Nvidia has been sitting on its ass for quite a while anyway by the way ( need we remind you that Pascal is Maxwell on a better manufacturing node?) Not to defend AMD but they have had inferior manufacturing nodes which holds them back both on the you market and on the CPU market, but that part could change and show that Nvidia isn't that far away and that in the long run both companies have similar perf with everything else being equal, but amd may have literally lost the silicon lottery with Glo Fo instead of TSMC these last few gens. You should try to have a bit more perspective to understand the current state of the market.)

 

Hold on there buddy, AMD was still flush with cash even after Intel's shenanigans. It was AMD's poor choice to buy ATi at a crazy ass price, with the r600 coming out as crap, and then Intel releasing its new Core line up that screwed AMD over.  Prior to the buy out of ATi, AMD had what 2 or 3 billion in cash or equivalent assets.  After they bought ATi, they had a debt of 5 billion and the same 2 or 3 billion in cash assets.

 

With the r600 screw up and subsequent HD3000 line, and Core 2 stomping the BD (phenom was still decent against Nehelam) line up, that is when AMD started going into their cash reserves.  After the HD4xxx line all the way up to the HD6xxx line the graphics cards were holding up the CPU side of things, then GCN was released. GCN also helped AMD stay afloat with its weak CPU's.  We don't have visibility on this, you know why?  Because after BD was released, AMD changed the way they filled SEC reports by combining GPU and CPU's under one umbrella.  All we can do is surmise it by looking at total bottom lines, where they made profits every now and then even with their bad CPU's.  AMD did this on purpose, they didn't want to show how bad their CPU's were doing.  It was to hide their weaknesses from investors.

 

 

Node doesn't matter, It was a choice AMD had to stick to because of the decisions they made in the past.

[mr moose]

7 minutes ago, Razor01 said:

  AMD did this on purpose, they didn't want to show how bad their CPU's were doing.  It was to hide their weaknesses from investors.

I don't know about on purpose, but it definitely made no difference to what the investors knew, their stocks fell and stayed low, and almost every economist on the planet was amazed they didn't go arse up for a long time.   

[laminutederire]

1 minute ago, Razor01 said:

Node doesn't matter, It was a choice AMD had to stick to because of the decisions they made in the past.

Node does matter, it's the only significant difference between Pascal and Maxwell and you have a huge gap between the two. There is a reason both CPU and GPU from Amd have trouble overclocking, the most probable being the manufacturing node they are using.

 

Having cash does not mean anything. You can contract long term debts by investing into r&d for instance, which won't affect your cash flow but will still be too much. Ultimately having cash is important, but it's nowhere near the only thing amd's management had to look at. Other things being their assets, their debts amount and when those are due, how much proper capital do they have and how much capital can be expected to flow in thanks to investors.

[Razor01]

25 minutes ago, laminutederire said:

Node does matter, it's the only significant difference between Pascal and Maxwell and you have a huge gap between the two. There is a reason both CPU and GPU from Amd have trouble overclocking, the most probable being the manufacturing node they are using.

 

Having cash does not mean anything. You can contract long term debts by investing into r&d for instance, which won't affect your cash flow but will still be too much. Ultimately having cash is important, but it's nowhere near the only thing amd's management had to look at. Other things being their assets, their debts amount and when those are due, how much proper capital do they have and how much capital can be expected to flow in thanks to investors.

I'm talking about node between AMD Polaris/Vega vs Pascal.  It doesn't matter, Polaris and Vega both on 14nm, only have the perf/watt of Maxwell

 

So bad products like BD didn't have a negative affect on AMD's cash flow?  They needed that cash to be used for R&D and they had the cash to do it but instead they decided to use VLSI automation design layouts in CPU design which all the engineers at AMD stated it was a bad choice.  They knew it was going to hurt performance via potential clock speeds and power consumption by 30% and still make a CPU that ate up more transistors.  That cash could have been used for R&D at that time.

 

All of these things is what is hurting AMD now, because of these moves, the CPU division got hurt, then to get that back on track, AMD couldn't spend money on the GPU division, the division that kept them from going bankrupt in the first place!  This is why the sr. engineers of the GPU division left mid way through GCN.  (put yourself in their shoes, if they helped create good products that helped a company stay afloat, and then the company cuts off much need moneys and resources to further push the envelope, you would probably feel slighted)  After all this, they got left behind with Maxwell, they had no back up plan,do you remember the 10 million dollar write off AMD took, think it was 10 million, for the failed 20 nm node?  Yeah they were going full blast with that node, without a back up of it if it failed.  nV didn't do that, they had a good back up they had Maxwell which was designed for 28nm.  Differences AMD couldn't budget in a 2nd team for another GPU.  All of this came from one thing AMD's bad choices in the past, crippled them.  To get more cash to stay afloat they spun off their foundries, by doing so hand cuffed them again, since now they had to use X amount of wafers from GF's.

[pas008]

so am I getting this right nvidia wants their cards to have separate sub brand naming and everyone is getting their panties in a bunch? and have access to their engineering of their designs to make a better product? and no one gets anything?

 

like they want to be excluded with msi gaming or asus rog or strix gigs aorus and be able to possibly look at what the vendors are doing with their product?

[mr moose]

8 minutes ago, pas008 said:

so am I getting this right nvidia wants their cards to have separate sub brand naming and everyone is getting their panties in a bunch? and have access to their engineering of their designs to make a better product? and no one gets anything?

 

like they want to be excluded with msi gaming or asus rog or strix gigs aorus and be able to possibly look at what the vendors are doing with their product?

The accusation is that nvidia will withhold essential support and products for partners who do not exclusively sell Nvidia under a "gaming" label, meaning they cannot sell an AMD product under any "gaming" label whilst a partner of the program.  

 

 

 

 

[pas008]

10 minutes ago, mr moose said:

The accusation is that nvidia will withhold essential support and products for partners who do not exclusively sell Nvidia under a "gaming" label, meaning they cannot sell an AMD product under any "gaming" label whilst a partner of the program.  

 

 

 

 

I didnt read any gaming label in article it states rog and nvidia wants their own tag to separate from amd

republic of gamers for amd

kingdom of gamers for nvida example lol

 

  would it mean to have your "Gaming Brand Aligned Exclusively With GeForce?" The example that will likely resonate best with HardOCP readers is the ASUS Republic of Gamers brand. I have no knowledge if ASUS is a GPP partner, I am simply using the ROG brand hypothetically. If ASUS is an NVIDIA GPP partner, and it wants to continue to use NVIDIA GPUs in its ROG branded video cards, computers, and laptops, it can no longer sell any other company's GPUs in ROG products. So if ASUS want to keep building NVIDIA-based ROG video cards, it can no longer sell AMD-based ROG video cards, and be a GPP partner.

 

[mr moose]

45 minutes ago, pas008 said:

I didnt read any gaming label in article it states rog and nvidia wants their own tag to separate from amd

republic of gamers for amd

kingdom of gamers for nvida example lol

 

  would it mean to have your "Gaming Brand Aligned Exclusively With GeForce?" The example that will likely resonate best with HardOCP readers is the ASUS Republic of Gamers brand. I have no knowledge if ASUS is a GPP partner, I am simply using the ROG brand hypothetically. If ASUS is an NVIDIA GPP partner, and it wants to continue to use NVIDIA GPUs in its ROG branded video cards, computers, and laptops, it can no longer sell any other company's GPUs in ROG products. So if ASUS want to keep building NVIDIA-based ROG video cards, it can no longer sell AMD-based ROG video cards, and be a GPP partner.

 

For GPU's most companies have only one gaming brand.  We have no first hand information about the "exclusiveness" of the gaming brand requirements as they are not presented in the article as a "quote from", they don't even talk about documents other than they have read some.  So all we have to go one is "AMD told us" and anonymous people told us some more stuff too.  

 

Quote

The crux of the issue with NVIDIA GPP comes down to a single requirement in order to be part of GPP. In order to have access to the GPP program, its partners must have its "Gaming Brand Aligned Exclusively With GeForce." I have read documents with this requirement spelled out on it.

 

[pas008]

Just now, mr moose said:

For GPU's most companies have only one gaming brand.  We have no first hand information about the "exclusiveness" of the gaming brand requirements as they are not presented in the article as a "quote from", they don't even talk about documents other than they have read some.  So all we have to go one is "AMD told us" and anonymous people told us some more stuff too.  

 

 

they should separate them though

2 completely different chips no matter the tier

about damn time

rog/auros/etc needs to be tied to the chip maker imho about time it happen

[mr moose]

9 minutes ago, pas008 said:

they should separate them though

2 completely different chips no matter the tier

about damn time

rog/auros/etc needs to be tied to the chip maker imho about time it happen

but why?   I can see why individuals might like such uniformity in their products, but companies like Asus and gigabyte like to cater for everyone.

 

EDIT: somewhere in all this someone suggested Asus could just start a branding called federation of gamers for AMD and ROG for Nvidia, Which would be nice but it would also increase the product cost a little as it requires more work/printing and marketing etc.  

[WMGroomAK]

8 minutes ago, pas008 said:

they should separate them though

2 completely different chips no matter the tier

about damn time

rog/auros/etc needs to be tied to the chip maker imho about time it happen

The added problem if you're Asus, MSI, Gigabytes, etc on this approach is that you already have brand name establishment with RoG, Aorus, Gaming X, and so on, so if you are splitting it up into different names then you have to re-establish brands on those new card series, which increases your costs.  Plus, I'm fairly sure that nVidia would not look kindly on a company for being relegated to a new naming tier over a well established one...  

[pas008]

3 minutes ago, mr moose said:

but why?   I can see why individuals might like such uniformity in their products, but companies like Asus and gigabyte like to cater for everyone.

 

EDIT: somewhere in all this someone suggested Asus could just start a branding called federation of gamers for AMD and ROG for Nvidia, Which would be nice but it would also increase the product cost incriminate as it requires more work/printing and marketing etc.  

eh

I think the damn naming needs to be simplified for each chip maker

rog intel/amd/nvidia products is ridiculous to me

the naming should be based on chipmaker imho

 

 

[pas008]

Just now, WMGroomAK said:

The added problem if you're Asus, MSI, Gigabytes, etc on this approach is that you already have brand name establishment with RoG, Aorus, Gaming X, and so on, so if you are splitting it up into different names then you have to re-establish brands on those new card series, which increases your costs.  Plus, I'm fairly sure that nVidia would not look kindly on a company for being relegated to a new naming tier over a well established one...  

then they dont have to be apart of the gpp simple as that

asus still can have rog strix same with any other company

who cares there needs to be a difference between chipmakers

 

if they were my chips I wouldnt want my product be associated with other companies under same sub brand 

[mr moose]

2 minutes ago, pas008 said:

eh

I think the damn naming needs to be simplified for each chip maker

rog intel/amd/nvidia products is ridiculous to me

the naming should be based on chipmaker imho

 

 

I think many argue they already do,  like most boxes have ion big letters down the side "1080 Ti" or "1050Ti"  and "RX580"  followed by ROG or G1 or whatever.

 

Unless I have missed something you are trying to convey?

[pas008]

Just now, mr moose said:

I think many argue they already do,  like most boxes have ion big letters down the side "1080 Ti" or "1050Ti"  and "RX580"  followed by ROG or G1 or whatever.

 

Unless I have missed something you are trying to convey?

see post above

[laminutederire]

2 hours ago, Razor01 said:

 

Except they are not because the processes aren't exactly the same. Look at Intel nodes for instance. It has been quite a while without node size change and architectural changes either, and yet newer CPU overclock better with better power consumption. That mostly comes down to the refinement of node manufacturing. That is what I'm talking about. I'm pretty confident that amd gpus with TSMC as a manufacturer would be at least 10% better, solely on overclock. But I can't prove it, but it remains something to think about.

 

Those product did have a negative impact, but the situation was pretty shaky even before.they invested in R&D with smaller returns  than deserved on the CPU side for a few generations, which led to them struggling a lot after one failed gen, since they didn't have enough sales before to get a cushion to fall back on.

 

The point was that the market did punish amd a lot every time and that was partially because of the mindshare of Nvidia or Intel that drove too much sales with inferior products more than once. And that mindshare is something that we should fight against because it will lead to Nvidia not be punished for any of their practices if they cheat, have a bad product, or have unreasonable prices for instance.

[pas008]

1 hour ago, laminutederire said:

Except they are not because the processes aren't exactly the same. Look at Intel nodes for instance. It has been quite a while without node size change and architectural changes either, and yet newer CPU overclock better with better power consumption. That mostly comes down to the refinement of node manufacturing. That is what I'm talking about. I'm pretty confident that amd gpus with TSMC as a manufacturer would be at least 10% better, solely on overclock. But I can't prove it, but it remains something to think about.

 

Those product did have a negative impact, but the situation was pretty shaky even before.they invested in R&D with smaller returns  than deserved on the CPU side for a few generations, which led to them struggling a lot after one failed gen, since they didn't have enough sales before to get a cushion to fall back on.

 

The point was that the market did punish amd a lot every time and that was partially because of the mindshare of Nvidia or Intel that drove too much sales with inferior products more than once. And that mindshare is something that we should fight against because it will lead to Nvidia not be punished for any of their practices if they cheat, have a bad product, or have unreasonable prices for instance.

there are documentation on difference between nodes from tsmc/intel/gf/samsung

they arent all created equal

intels 14nm++ is like others 10nm

think that is why intel is having issues with 10nm, dont quote me on that but yeah,

[WMGroomAK]

10 minutes ago, pas008 said:

if they were my chips I wouldnt want my product be associated with other companies under same sub brand

The problem that you begin to run into with this is that it begins to skate really close to the whole Intel v AMD case from about a decade ago...  What's to stop Intel from telling Dell that they can only advertise their processors under a certain computer name like Alienware? 

[pas008]

1 minute ago, WMGroomAK said:

The problem that you begin to run into with this is that it begins to skate really close to the whole Intel v AMD case from about a decade ago...  What's to stop Intel from telling Dell that they can only advertise their processors under a certain computer name like Alienware? 

how is that even remotely close

 

each chipmakers products should have names not associated with a competitor

simple as that

[Razor01]

41 minutes ago, laminutederire said:

Except they are not because the processes aren't exactly the same. Look at Intel nodes for instance. It has been quite a while without node size change and architectural changes either, and yet newer CPU overclock better with better power consumption. That mostly comes down to the refinement of node manufacturing. That is what I'm talking about. I'm pretty confident that amd gpus with TSMC as a manufacturer would be at least 10% better, solely on overclock. But I can't prove it, but it remains something to think about.

 

Those product did have a negative impact, but the situation was pretty shaky even before.they invested in R&D with smaller returns  than deserved on the CPU side for a few generations, which led to them struggling a lot after one failed gen, since they didn't have enough sales before to get a cushion to fall back on.

 

The point was that the market did punish amd a lot every time and that was partially because of the mindshare of Nvidia or Intel that drove too much sales with inferior products more than once. And that mindshare is something that we should fight against because it will lead to Nvidia not be punished for any of their practices if they cheat, have a bad product, or have unreasonable prices for instance.

 

 

The node doesn't matter, AMD is stuck with GF so they had to use GF's 14nm, they had no choice, so we don't even need to discuss it here.

 

The problem wasn't the node either look at Zen why is it capable of so much more Mhz than Polaris?  Its designed for it.  Yeah I can see 10 % difference maybe because of design sure, but not purely due to clock increases at the same voltage, more voltage, which is something Zen was capable of handling and Polaris wasn't.  Polaris peaked out at 1.3 V, Zen went to 1.34, guess what GF 1050 and ti went to, yeah 1.2ish......  Now we see three different architecture, two GPU's and one CPU, all having different tolerances to voltage with the same node.  Oh even better 1.2 volts was the max voltage of gtx 1050 and ti TSMC 16 nm too.  Coincidence, nope, read next para.

 

ASIC design dictates what voltage and mhz at what power levels will reach.  Nodes have their max voltage limits but that is not the basis of chip design.  These GPU designers are well aware of what voltage, what mhz, and at what power these chips are coming back with based on the node they choose well before the chip even goes into risk production.

 

The problem wasn't with the node, its GCN.

 

GCN from Day one was a power hungry chip, akin to nV's Fermi.  When a chip is designed for all in one purposes, its going to have silicon that isn't best suited for individual tasks.  This is why nV started different chip designs for different market segments. This is what Raja was talking about with AMD's next move with Navi,  this is what scale-able means, scale-able across multiple market needs.  Not scale-able as in multi chip designs.

[pas008]

8 minutes ago, Razor01 said:

 

 

The node doesn't matter, AMD is stuck with GF so they had to use GF's 14nm, they had no choice, so we don't even need to discuss it here.

 

The problem wasn't the node either look at Zen why is it capable of so much more Mhz than Polaris?  Its designed for it.  Yeah I can see 10 % difference maybe because of design sure, but not purely due to clock increases at the same voltage, more voltage, which is something Zen was capable of handling and Polaris wasn't.  Polaris peaked out at 1.3 V, Zen went to 1.34, guess what GF 1050 and ti went to, yeah 1.2ish......  Now we see three different architecture, two GPU's and one CPU, all having different tolerances to voltage with the same node.  Oh even better 1.2 volts was the max voltage of gtx 1050 and ti TSMC 16 nm too.  Coincidence, nope, read next para.

 

ASIC design dictates what voltage and mhz at what power levels will reach.  Nodes have their max voltage limits but that is not the basis of chip design.  These GPU designers are well aware of what voltage, what mhz, and at what power these chips are coming back with based on the node they choose well before the chip even goes into risk production.

 

The problem wasn't with the node, its GCN.

 

GCN from Day one was a power hungry chip, akin to nV's Fermi.  When a chip is designed for all in one purposes, its going to have silicon that isn't best suited for individual tasks.  This is why nV started different chip designs for different market segments. This is what Raja was talking about with AMD's next move with Navi,  this is what scale-able means, scale-able across multiple market needs.  Not scale-able as in multi chip designs.

what? many times here

I dont want to start bolding but what?