Rumor: AMD gets priority advantage to Hynix HBM2

[Genx87]

Has nothing to do with if they can or cannot. They would have to do it without infringing on AMD's patent, that or license the patent from them.

Im sure they can do it without infringing on their patents. It isnt like Nvidia hasn't had years to work on it. Or maybe they will just license it. Who knows. The point is an MC wont be an issue for Nvidia and HBM2. They taped out Pascal a month and a half ago.

 

[ocre]

Not true and definitely not better that way.

eg. Say witcher 3 devs used tressfx instead of hairworks. That experience would last them for their next game, their expansions etc. They could freely use that tech anywhere, anytime. Now they have to go to nvidia for permission everytime and they don't even have freedom with the tech.

The same with Physx. Havok is better, bullet is better. Tons of games have solid physics without physx. Same with shadow tech, PCSS does not look better than softshadows and performs worse than AMDs CHS and Softest shadow setting

http://www.hardocp.com/article/2015...video_card_performance_preview/6#.Vae_WvlVhBc

In reality its less performance for inferior tech.

The contrast here is that one is genuine hardware innovation and a huge boon for gamers. The other is just bloatware pretending to be added value.

Why can't NV play nice like AMD, they develop the features, make it open, push forward great optimized features for all to enjoy?

As said, only one scenario benefits PC gamers. GW doesn't even benefit NV gamers, because its horribly optimized. A frequently complaint on the steam forum is from NV owners why they can't get great smooth performance with it enabled.

If NV make it open, devs would be better able to optimize it themselves for all hardware. So instead of a major performance hit, you could have a very light hit like TressFX. Now all gamers can enjoy.

Yet no one even tried to use tress effect on Witcher. You guys have some huge imaginations. There is a difference between getting it done and dreaming.
Gameworks is more than hair, and hair works is much more than tress effect on Lora crafts hair.

Check out flameworks on you tube, you guys specifically might denounce it here, but deep down you will secretly know it is pretty freakn amassing.

Gameworks is so much more than tress effect, you guys go as far as dissing and downing everything nvidia ever done. That probably will never change. But havok is better than physX, what do u base that on? How one developer chooses to implement, that's all on them. Also, batman AK was plagued with issues. It wasn't Gameworks at all, it was the developers choice and method in porting it. The developer took full responsibility but you will continue to spread misinformation like your life depends on it.

tress effect, which would cover a fraction of one effect in Gameworks, is here and open for whatever developer wants to invest in and use. But no one is spending that extra time and money for PC gaming. No one.
Get real guys. Wake up and quit spouting fantasies. Your imaginations are larger than your reasoning....that's my opinion on your total spin tactics

 

[railven]

This post wins the thread.

As others have said, if AMD has first exclusivety, then it will be short-lived. If Nvidia is many months ahead of AMD's next-gen launch, then I don't see how Hynix sits on their hands losing money, unless AMD signed some sort of agreement with Hynix similar to their GloFo write-downs where they pay for the chips whether or not the chips are needed.

Totally forgot about that deal AMD signed. Didn't they lose millions on it? Sounds about right then. Good catch, we know AMD's plan then.

 

[Keysplayr]

I don't think it's a viable strategy to stick with GDDR5 because of the much improved efficiency HBM brings. Nvidia dominates dGPU notebook sales because of it's superior efficiency. Perhaps only the flagship Pascal die uses HBM, but I think after Pascal all chips will be using HBM, even the low end ones. Being able to stay comfortably ahead of Intel's Iris Pro is important for Nvidia as well.

Agreed. I just meant that if they had to continue using GDDR5 due to the AMD lockout that may or may not happen.

 

[Headfoot]

For GPU patents they do. But memory is not part of the GPU. Its a another component that can be used on GPUs.

Distinction without a difference. Unless you can point to a reason their cross licensing wouldnt include that, I very very very highly doubt it. Cross licensing agreements far more likely than not are for all patents both ways. Sometimes plus dollars/royalties if one side has a weaker portfolio

if AMD has exclusivity on hbm2 against nVidia its not because of patents. If there is any legal exclusivity, it would be that they cut a deal with Hynix in a contract.

My bet is they were just first in line so they got their orders in before anyone else could. I doubt its a grand conspiracy or legal maneuvering. Good old case of First In First Out.

 

[Stuka87]

Distinction without a difference. Unless you can point to a reason their cross licensing wouldnt include that, I very very very highly doubt it. Cross licensing agreements far more likely than not are for all patents both ways. Sometimes plus dollars/royalties if one side has a weaker portfolio

if AMD has exclusivity on hbm2 against nVidia its not because of patents. If there is any legal exclusivity, it would be that they cut a deal with Hynix in a contract.

My bet is they were just first in line so they got their orders in before anyone else could. I doubt its a grand conspiracy or legal maneuvering. Good old case of First In First Out.

It is never for all patents. Patents are listed out in the agreements. AMD has tons of IP that is in no way licensed to nVidia as part of a GPU cross-license agreement, most of it being CPU related. There may also be one for CPU related IP, but its not all in one inclusive agreement.

I am not saying AMD's HMB2 exclusivity is due to the patents, definitely not the case. A second discussion here is that HBM has been ratified as a standard, but that does not mean the memory control that is required for it has been (And as far as we can tell, it is not). So the memory controller would need to be licensed on its own outside of any HBM licensing.

 

[tviceman]

Totally forgot about that deal AMD signed. Didn't they lose millions on it? Sounds about right then. Good catch, we know AMD's plan then.

Only now, I don't think AMD can afford to buy chips / wafers they won't use. They're running out of cash reserves and, at their current pace, will be cash insolvent by Q3/Q4 2017. I'm guessing if there is any sort of agreement, it's that AMD has first option to buy or decline as soon as HBM2 is commercially available, but since they can't afford purchases too far in advance, if Nvidia is ready 12+ weeks before AMD with HBM2 then AMD will forego their first dibs option.


Just my speculation on if an agreement exists. I doubt it does. Nvidia has been working on HBM equipped graphics chips for some time now so I'm sure Hynix will ramp up supply quite quickly.

 

[Headfoot]

It is never for all patents. Patents are listed out in the agreements. AMD has tons of IP that is in no way licensed to nVidia as part of a GPU cross-license agreement, most of it being CPU related. There may also be one for CPU related IP, but its not all in one inclusive agreement.

I am not saying AMD's HMB2 exclusivity is due to the patents, definitely not the case. A second discussion here is that HBM has been ratified as a standard, but that does not mean the memory control that is required for it has been (And as far as we can tell, it is not). So the memory controller would need to be licensed on its own outside of any HBM licensing.

There's no way you can know that unless you have access to a copy of their agreement. And you are 100% wrong on "it's never for all patents." Are you a patent attorney?

http://newsroom.cisco.com/release/1342531/Cisco-and-Samsung-Enter-Into-Patent-Cross-License-Agree_2 -< an example of Cisco and Samsung cross licensing ALL of their patents, and all their patents to come for 10 years.

Total cross license agreements are extremely popular among tech Fortune 500s, and this is highly common knowledge in the intellectual property field. They just want blanket immunity from suit from a major competitor, and so does that competitor. So they absolutely do NOT list every patent because that list would become insufficient the moment a new patent is issued for either side. Why would you amend the agreement constantly when you can just cross license every patent, and every future patent. Exactly how Cisco and Samsung did in the example above

IBM and Samsung famously did an all disk drive tech cross license, so its also popular to limit it to one tech category too where your business' only partially overlap.

But where the industries of the players are similar its much easier and way less likely to provoke litigation to do a full portfolio cross license. And the only reason you do a cross license deal is to avoid litigation. I highly suspect this is what AMD and nVidia have. Many GPU and CPU patents are going to be EE patents which cover any use of that sort of transistor/layout/method etc. Many of claims cover both CPU and GPU applications, intentionally. I'm sure both AMD and nVidia have CPU specific patents as well as GPU specific patents. I know there are lots of graphics method patents which could be specific to GPU to some degree although you can execute graphics code slowly on a CPU. They both have thousands of patents each with a dozen or two claims and its a monumental task to sort through those compared to a full cross license.

Pharmaceutical patents are usually the ones that are listed on very a specific basis, generally speaking.

bottom line, its plausible they have a tech specific list-based cross licensing deal, but its highly unlikely. And if they do have it split down by tech category I'm sure it would be based on ego reasons and not business ones.

 

[Azix]

yeah. People did criticize AMD on not trying to get them to use tressfx earlier. Dropped the ball. My guess is nvidia took the initiative rather than the devs asking for it. Based on whats in the game it's clear the witcher devs had their own hair stuff setup already.

Yes gameworks is more than hairworks. fireworks looks cool in demo, but none of nvidias demos interest me till I see it in game. A demo won't tell me I won't lose 20-35 fps by turning on the one effect. You can understand none of these more significant effects can currently be run together without killing performance completely. There are always alternatives. The thing about nvidia is they are giving a name to every individual effect. They even call something faceworks. Fire, smoke water can all be done in a physics engine without giving them each a special name. They call another waveworks. come on.

We are not equating tressfx to gameworks, just hairworks. both hairworks and tressfx go beyond just hair. As far as spending time.. .depends. They spent a lot of time building in hair and fur before hairworks got in there. Considering tressfx works on consoles too, that might have saved them quite some time.

Yet no one even tried to use tress effect on Witcher. You guys have some huge imaginations. There is a difference between getting it done and dreaming.
Gameworks is more than hair, and hair works is much more than tress effect on Lora crafts hair.

Check out flameworks on you tube, you guys specifically might denounce it here, but deep down you will secretly know it is pretty freakn amassing.

Gameworks is so much more than tress effect, you guys go as far as dissing and downing everything nvidia ever done. That probably will never change. But havok is better than physX, what do u base that on? How one developer chooses to implement, that's all on them. Also, batman AK was plagued with issues. It wasn't Gameworks at all, it was the developers choice and method in porting it. The developer took full responsibility but you will continue to spread misinformation like your life depends on it.

tress effect, which would cover a fraction of one effect in Gameworks, is here and open for whatever developer wants to invest in and use. But no one is spending that extra time and money for PC gaming. No one.
Get real guys. Wake up and quit spouting fantasies. Your imaginations are larger than your reasoning....that's my opinion on your total spin tactics

 

[Silverforce11]

@ocre
Way to miss the point completely.

GW would be better for gamers & developers if it was open source, modifiable, optimizable by others besides NV. We as gamers can enjoy all those effects, but at a less performance hit.

Just a simple example for you, as a gamedev, I would optimize HairWorks and find x16 tessellation offer the best IQ vs performance. More than that, has too low an impact on IQ or none at all, but costs a lot of performance. Or, I would be able to modify the code and offer various levels for users to select, not just one default x64.

We've seen only a few GW features in games so far, mostly HBAO, DoF and HairWorks. The other feature demos look fantastic, but if I had a guess why they aren't used more, its probably because they are unoptimized and tank performance hard. HairWorks cut performance in half, add some more GW features and we can all enjoy a slideshow, on top of the line GPUs.

 

[Abwx]

Whatever AMD does in respect of HBM2, exclusivity or not, i would just invite people to imagine what would have happened if the tables were turned...

 

[crisium]

If the tables were completely turned and Nvidia had poor marketshare and AMD did Gameworks then people would feel the same way. AMD having a leg up in HBM2 is good for the industry because they are approaching 20% marketshare. This would mean AMD would lessen the gap in efficiency, begin selling more, and Nvidia prices would drop. Those that don't want as close to 50/50 marketshare as possible are blinded by fanboyism (or have an invested interest, fair enough).

 

[Keysplayr]

If the tables were completely turned and Nvidia had poor marketshare and AMD did Gameworks then people would feel the same way. AMD having a leg up in HBM2 is good for the industry because they are approaching 20% marketshare. This would mean AMD would lessen the gap in efficiency, begin selling more, and Nvidia prices would drop. Those that don't want as close to 50/50 marketshare as possible are blinded by fanboyism (or have an invested interest, fair enough).

Not really 100% true, that last comment. Things would be 50/50 if there were two companies that fought tooth and nail for it. As it turns out, we have one company that does that, and another that is 95% (made up statistic as 95% of all are) reactive and only occasionally does something cool. HBM is cool.
So, the market share is what it is for a reason. And no folks, it's not JUST Nvidia's superior marketing of inferior products. Always do one of these when I hear that excuse. Good marketing is just one part of a well oiled company that can continually pump out industry leading products. Intel would be another. Just saying. Apple.

 

[AnandThenMan]

You are not understanding what marketing and brand awareness does. It creates a snowball effect, the better your brand is perceived the better it sells and the more cash you have to make better products. Put very simply, marketing is the foundation you build on, especially in the Internet age.

 

[Keysplayr]

You are not understanding what marketing and brand awareness does. It creates a snowball effect, the better your brand is perceived the better it sells and the more cash you have to make better products. Put very simply, marketing is the foundation you build on, especially in the Internet age.

Marketing is an important part of any business, but I was under the impression you had to have an actual good product in order for the marketing to sustain itself in the long term?

You can have a crappy product, and the best marketing in the world and sell a ton of that crappy product. But don't expect to do it twice. Nvidia had/has great marketing of great products. It goes hand in hand man.

It just gets old and tired of hearing what is essentially saying "Nvidia has the most market share because of it's marketing. Not because it has as good or better products. People buy Nvidia because they don't know any better"

I mean sheesh the bs can get thicker than a slug trail.

Anyway, sorry for the OT. I think whatever the methods of getting there, Nvidia will have HBM2 on Pascal based GPUs. Whether or not they have to pay licensing fees for the interposer tech, or create their own, who knows. Whatever way works. They'll have HBM2 on their GPUs.

 

[Abwx]

You can have a crappy product, and the best marketing in the world and sell a ton of that crappy product. But don't expect to do it twice. Nvidia had/has great marketing of great products.

Then 30 millions chips that busted in laptops should had got them completely bankrupted, there s no other exemple of such a terrible quality delivered in such a large scale....

Great marketing if you want but as far as product quality they have the worst record of all industry, i guess that tons of marketing were needed after this failure...

 

[Keysplayr]

Then 30 millions chips that busted in laptops should had got them completely bankrupted, there s no other exemple of such a terrible quality delivered in such a large scale....

Great marketing if you want but as far as product quality they have the worst record of all industry, i guess that tons of marketing were needed after this failure...

That's off the deep end bud. Insane exaggeration as usual.

Every company has it's problems from time to time. But you know what, it wasn't the GPUs that failed. It was the solder. I still have an 8600GT working fine in a laptop here. And 30 millions? Sound like a huge number.

 

[Abwx]

That's off the deep end bud. Insane exaggeration as usual.

Every company has it's problems from time to time. But you know what, it wasn't the GPUs that failed. It was the solder. I still have an 8600GT working fine in a laptop here. And 30 millions? Sound like a huge number.

That s a conservative number actualy, that said i wont insist on this case but it s telling that what matters is the perception, otherwise Intel for exemple wouldnt had retained the lead in the P4 era.

I think that there s numerous instances where the marketshares are not really representative of the product s real qualities or lack of, and in this respect this is surely what AMD lacks, namely an agressive marketing, not that i support such move but they have to adapt to their competitor s methods...

 

[AnandThenMan]

In my view AMD has moved the industry forward much more than Nvidia, NV tends to wait things out and capitalize (literally) on existing tech. AMD brought us the first DX9 GPU, first DX11 as well, AMD64 (transformed the industry), GDDR3, instrumental in GDDR5, Mantle (another industry changer), HBM/HBM2.

Nvidia has what to their name, SLI, PhysX, .... ??? BTW bumpgate cost Nvidia $475 million, and it would have been much higher if they had replaced the defective GPUs with ones that were not defective. As for some out there still working well sure, you sell 30+ million of anything and some of them are bound to keep working. I'm sure there are a few Chevy Vega's still on the road with the original engine.

Back to the basic topic here, if AMD gets exclusive or preferred access to HBM2 then good on them they should seeing the company spent 7 years of research and development to bring it to market.

 

[iiiankiii]

That's off the deep end bud. Insane exaggeration as usual.

Every company has it's problems from time to time. But you know what, it wasn't the GPUs that failed. It was the solder. I still have an 8600GT working fine in a laptop here. And 30 millions? Sound like a huge number.

It actually upwards of 70 million chips that were prone to failure due to those weak solder points. Note, I said prone to failure. G84, G86 and G92 GPUs for desktop AND laptop all had weak solder points. Desktop GPUs were less prone to solder points cracking because it had much better cooling. The laptop versions weren't as lucky for obvious thermal reasons.

We don't know exactly how many GPUs failed because of those weak soldering point. I can say that I had atleast 3 laptop failures due to bumpgate. 2 Dells with 8400m and an HP with an 8600m GT. On the plus side, the replacement I got from Dell with an 8600m GT still works to this day (although it hasn't been touched in years).

I also had a desktop 8800GT failed on me. It was "baked" back to life a few times before I finally gave up on it.

 

[Abwx]

In my view AMD has moved the industry forward much more than Nvidia, NV tends to wait things out and capitalize (literally) on existing tech. AMD brought us the first DX9 GPU, first DX11 as well, AMD64 (transformed the industry), GDDR3, instrumental in GDDR5, Mantle (another industry changer), HBM/HBM2.

They moved the industry because historically they are a pure enginering firm with about no marketing other than Jerry sanders himself, that s why once he retired they had trouble replacing his leadership.

Back to the basic topic here, if AMD gets exclusive or preferred access to HBM2 then good on them they should seeing the company spent 7 years of research and development to bring it to market.

According to a SA member who has some knowledge of the timing HBM2 wont be here untill second half of 2016, so the noise that Nvidia could release such a GPU in H1 2016 are just willfull thoughts, probability is that they ll have to wait a quarter or two after AMD initiate the first shippements.

 

[HurleyBird]

So, the market share is what it is for a reason. And no folks, it's not JUST Nvidia's superior marketing of inferior products. Always do one of these

Do you have an argument to go with that assertion?

 

[Rvenger]

No, he doesn't because I am putting a stop to this. This thread is going nowhere. It always has to be an AMD vs. Nvidia war and I am not even entertaining it anymore.

-Rvenger