[Kitguru ] AMD’s partners cannot get enough Radeon R9 Fury graphics cards –

[Sabrewings]

http://techreport.com/review/28294/amd-high-bandwidth-memory-explained



AMD created HBM. You can find patents in public records online, but it'll take some searching to find.

JEDEC is all about open standards. If it's JEDEC approved, either they don't have a patent for it or they signed an agreement that it's free for anyone to develop without fear of litigation.

JEDEC's adoption of open industry standards (i.e., standards that permit any and all interested companies to freely manufacture in compliance with adopted standards) serves several vital functions for the advancement of electronic technologies. First and foremost, such standards allow for interoperability between different electrical components. JEDEC standards do not protect members from normal patent obligations. The designated representatives of JEDEC member companies are required to disclose patents and patent applications of which they personally are aware (assuming that this information is not considered proprietary). JEDEC patent policy requires that standards found to contain patents, whose owners will not sign a standard JEDEC patent letter, be withdrawn. Thus the penalty for a failure to disclose patents is retraction of the standard. Typically, standards will not be adopted to cover technology that will be subject to patent protection. In rare circumstances, standards covered by a patent may be adopted, but only on the understanding that the patent owner will not enforce such patent rights or, at a minimum, that the patent owner will provide a reasonable and non-discriminatory license to the patented technology.

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

 

[Sabrewings]

It looks to me like it is for the memory controller used with HBM although it is written for a more general application as well. From the patent:

The basic abstract has nothing to do with how HBM is used which is where a lot of litigation can be fought. It's a terrible patent if it is for HBM, which it doesn't appear to be regardless of using chip stacks. It talks about reliability (parity) data being stored more than it does chip stacks.

An integrated circuit includes a memory having an address space and a memory controller coupled to the memory for accessing the address space in response to received memory accesses. The memory controller further accesses a plurality of data elements in a first portion of the address space, and reliability data corresponding to the plurality of data elements in a second portion of the address space.

 

[Termie]

yeah it doesn't work in my searches either. only if i go to this direct link then it works:

http://www.newegg.com/Product/Product.aspx?Item=N82E16814202156

although it is available the the main video card section:
http://www.newegg.com/Video-Cards-Video-Devices/Category/ID-38

FYI, the Fury Tri-X sold out at 12:23pm Pacific Time. I was watching it pretty closely.

 

[Elixer]

We will see at the steam survey (If the Fury series can get over the minimum limit). Because like the marketshare it disagrees with the notions that people got on this forum about how things actually sells. Entire R9 200 series for once.

Steam's survey is 100% voluntary, so, the results are always skewed.
So, the survey doesn't tell the whole story.
Intel HD Graphics 4000 is #1, and Intel HD Graphics 3000 is #2, with the 970 coming in 3rd.

What does this tell us? Only that the vast majority of the people who submitted the survey have Intel integrated gfx.

And if you look at sales, you get back a 'no kidding!' response.

 

[Hitman928]

The basic abstract has nothing to do with how HBM is used which is where a lot of litigation can be fought. It's a terrible patent if it is for HBM, which it doesn't appear to be regardless of using chip stacks. It talks about reliability (parity) data being stored more than it does chip stacks.

I didn't say it was a patent for HBM. It looks like a patent for their memory controller integrated into HBM and how to deal with memory access and errors in a stacked die. I would be surprised if that is part of the HBM JEDEC standard.

 

[Sabrewings]

I didn't say it was a patent for HBM. It looks like a patent for their memory controller integrated into HBM and how to deal with memory access and errors in a stacked die. I would be surprised if that is part of the HBM JEDEC standard.

In either case, it is definitely not relevant to Nvidia's inability to use HBM by patent blocking (which isn't possible due to it being a JEDEC standard, regardless of who developed it).

Once you submit to JEDEC and it's approved you can't enforce a patent on it.

It could only be done through Hynix unfairly blocking sales to Nvidia, but if so other memory companies would be jumping to produce it and make big bucks off Nvidia.

 

[Elixer]

Once you submit to JEDEC and it's approved you can't enforce a patent on it.

Wrong.
Look at what RAMBUS did.
So, yes, AMD could have patents on lots of things, and, make it very difficult for nvidia to do what AMD has done. Not so much for Intel, since they share technology.

 

[monstercameron]

In either case, it is definitely not relevant to Nvidia's inability to use HBM by patent blocking (which isn't possible due to it being a JEDEC standard, regardless of who developed it).

Once you submit to JEDEC and it's approved you can't enforce a patent on it.

It could only be done through Hynix unfairly blocking sales to Nvidia, but if so other memory companies would be jumping to produce it and make big bucks off Nvidia.

maybe they are have to do frand once jedec have accepted the standard.
https://en.wikipedia.org/wiki/Reasonable_and_non-discriminatory_licensing

 

[Kenmitch]

In either case, it is definitely not relevant to Nvidia's inability to use HBM by patent blocking (which isn't possible due to it being a JEDEC standard, regardless of who developed it).

Once you submit to JEDEC and it's approved you can't enforce a patent on it.

It could only be done through Hynix unfairly blocking sales to Nvidia, but if so other memory companies would be jumping to produce it and make big bucks off Nvidia.

http://www.freshpatents.com/Gabriel-H-Loh-Bellevue-invdxl.php

Guy looks to be inventive. On phone so didn't look thru it much. No vested interest for me so could care less either way. Should be in it's own thread anyways.

Looks like fury X pops in and out of availability. Every one made looks to be selling at least. Supply #'s are the unknown variable currently.

 

[Sabrewings]

Wrong.
Look at what RAMBUS did.
So, yes, AMD could have patents on lots of things, and, make it very difficult for nvidia to do what AMD has done. Not so much for Intel, since they share technology.

What exactly did they do that is analogous to what's going on? They flirted with JEDEC for years but ended up pulling away because they didn't agree with the patent rules. Do you have an example of something they had a patent on, that was approved by JEDEC, and that they were allowed to enforce that patent on?

FYI:

n the early 1990s, Rambus was invited to join the JEDEC. Rambus had been trying to interest memory manufacturers in licensing their proprietary memory interface, and numerous companies had signed non-disclosure agreements to view Rambus' technical data. During the later Infineon v. Rambus trial, Infineon memos from a meeting with representatives of other manufacturers surfaced, including the line "[O]ne day all computers will be built this way, but hopefully without the royalties going to Rambus", and continuing with a strategy discussion for reducing or eliminating royalties to be paid to Rambus. As Rambus continued its participation in JEDEC, it became apparent that they were not prepared to agree to JEDEC's patent policy requiring owners of patents included in a standard to agree to license that technology under terms that are "reasonable and non-discriminatory",[7] and Rambus withdrew from the organization in 1995. Memos from Rambus at that time showed they were tailoring new patent applications to cover features of SDRAM being discussed, which were public knowledge (JEDEC meetings are not secret) and perfectly legal for patent owners who have patented underlying innovations, but were seen as evidence of bad faith by the jury in the first Infineon v. Rambus trial. The Court of Appeals for the Federal Circuit (CAFC) rejected this theory of bad faith in its decision overturning the fraud conviction Infineon achieved in the first trial (see below).

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

If they don't allow their standard to be licensed in a fair and non-discriminatory way, it's not allowed to be a standard. So, if Hynix/AMD (assuming they have an approved patent, as an application does not equal a patent) do not allow other memory manufacturers to produce HBM, the standard is revoked. This would allow a similar but different enough technology for another patent to become the standard and Hynix/AMD would be up a creek.

 

[myocardia]

If it helps the discussion, retailers here that I've talked to, said they got a single shipment of Fury X which was sold out instantly but now their distributors cannot even confirm an ETA on the next shipments. A few of them halted pre-orders completely because they have a lot on the list and they have no clue when its going to be in stock.

That would imply that the story in the OP is correct, then. Not that any of us should be surprised. It is extremely normal to have delays, when introducing any new technology. The "supply is lower than demand" problem is the most normal one of them all to have, with any new technology. It honestly would have been a much bigger surprise to me if AMD had pulled this off without any delays or supply problems, since I doubt that either Intel or nVidia could have pulled it off without those same problems.

 

[Elixer]

If they don't allow their standard to be licensed in a fair and non-discriminatory way, it's not allowed to be a standard. So, if Hynix/AMD (assuming they have an approved patent, as an application does not equal a patent) do not allow other memory manufacturers to produce HBM, the standard is revoked. This would allow a similar but different enough technology for another patent to become the standard and Hynix/AMD would be up a creek.

Who said anything about not allowing others to make HBM, that is only part of the issue here. The other part is how to connect HBM to your package, and that is where AMD has patents that can make it difficult for anyone else to follow what they have done.
That is all I was saying.

For the availability of the Fury line of cards...
http://www.nowinstock.net/computers/videocards/amd/r9fury/
http://www.nowinstock.net/computers/videocards/amd/r9furyx/

 

[Sabrewings]

Who said anything about not allowing others to make HBM, that is only part of the issue here. The other part is how to connect HBM to your package, and that is where AMD has patents that can make it difficult for anyone else to follow what they have done.

Then they're still breaking the rules of the standard. You can't have an approved JEDEC standard but unfairly enforce a patent on the way it is implemented. It defeats the purpose of it being a standard and JEDEC knows that.

Edit: BTW, in JESD235 (HBM JEDEC standard), the memory logic base die is not required to implement HBM.

The DRAM vendor may choose to require an optional interface die that sits at the bottom of the stack and
provides signal redistribution and other functions. The vendor may choose to implement many of the logic
functions typically found on DRAM die on this logic die. This standard does not explicitly require nor
prohibit such a solution.

 

[Eymar]

So let me get this straight. When the bitcoin craze was at it's full height, AMD couldn't deliver because........???? When Nvidia came out with Titan AMD couldn't deliver because.......???? Now that AMD is competitive again at the high end on most metrics they can't deliver because.......????

Haha, because.....they literally can't deliver. AMD is not good with launches, but they do have competitive high end product line with Fury. Hopefully, they did launch with a higher MSRP for Fury as they know initial inventory will sell out and gives them some wiggle room to lower prices come back to school time. In regards to inventory, aren't these cards shipped by sea freight so each vendor will have at least one week or more between shipments (travel and port loading)? I wonder that's why Nvidia always had Titan X in stock as they might of used air freight?

 

[Silverforce11]

And AMD is going to have HBM for a year before Nvidia. If it's an open standard other companies will be jumping on it to produce and cash in. We'll see when (if) Nvidia updates their road map.

It's not an open standard as yet, since only SK Hynix produces HBM and only they so far have production rights to HBM2. Thus, this suggests AMD got SK Hynix onboard with HBM with the rights going to one producer only. There is probably a timing involve with other producers, so AMD may have struck a deal, SK Hynix gets to be sole producer, AMD gets to be the sole recipient until X time.

If true, Pascal is going to need a redesign for GDDR5.

ps. It'll be "open" when like GDDR5, there's many producers of chips. But also, the chip may be open but certainly the integration and the IMC stack may be covered by patents and thus, it's proprietary until NV develop their own method to utilize HBM chips. If AMD was smart, they would fubar NV, stop giving them freebies like GDDR3, GDDR5 and now HBM2.

pss. AMD did deliver during Titan-era with the R290/X, shortly after, marketshare went up, on the trajectory to 40/60 which is quite good compared to the 20/80 in recent times. But due to pricing, I doubt Fury X/Pro will move marketshare much.

 

[Sabrewings]

It's not an open standard as yet, since only SK Hynix produces HBM and only they so far have production rights to HBM2

Yes, it is. JEDEC Standard 235

https://www.jedec.org/standards-documents/docs/jesd235

Big bold letters, first page of the document. "JEDEC Standard - High Bandwidth Memory DRAM"

It's as much a standard as any other JEDEC standard and I encourage you to prove otherwise. As such, anyone can produce it within the FRAND limitations. They can't simply say "no" and keep it to themselves.

 

[svenge]

Steam's survey is 100% voluntary, so, the results are always skewed.
So, the survey doesn't tell the whole story.
Intel HD Graphics 4000 is #1, and Intel HD Graphics 3000 is #2, with the 970 coming in 3rd.

What does this tell us? Only that the vast majority of the people who submitted the survey have Intel integrated gfx.

While the absolute number of graphics card sales can't be derived from Steam's survey, the proportional ratio of sales for "Card A" and "Card B" can be reasonably inferred. That is to say that given the percentage of respondants with GTX 970 cards (2.84%) and the R9 200 Series as a whole (0.85%), it is quite likely that the GTX 970 probably sold somewhere between 3 - 3.5x as many units as all the R9 200 Series models combined, even if we don't know the actual number sold.

That is unless you can think of some statistically valid reason that NVIDIA owners would be more willing to engage in Steam's "push" survey than AMD owners, which I find highly unlikely...

 

[Sabrewings]

But also, the chip may be open but certainly the integration and the IMC stack may be covered by patents and thus, it's proprietary until NV develop their own method to utilize HBM chips.

BTW, the standard has been out for nearly two years. You really believe Nvidia hasn't had time to work out a controller method for it of their own?

If AMD was smart, they would fubar NV, stop giving them freebies like GDDR3, GDDR5 and now HBM2.

They can't. If AMD wanted to do that, they should've stayed away from JEDEC. Nvidia is a member of JEDEC, so if you read the rules they have to comply by allowing RAND licensing.

https://www.jedec.org/about-jedec/patent-policy

 

[Silverforce11]

BTW, the standard has been out for nearly two years. You really believe Nvidia hasn't had time to work out a controller method for it of their own?

Maybe they were hoping AMD will be as gracious as the past memory tech and share it freely (from the same company who says GameWorks is their IP, they won't share it!). But then the patent showed up.. oops.

 

[Sabrewings]

Maybe they were hoping AMD will be as gracious as the past memory tech and share it freely (from the same company who says GameWorks is their IP, they won't share it!). But then the patent showed up.. oops.

They'd be stupid to sit on their laurels for years with a standard out and just hope "AMD is going to give it to us."

If Nvidia is anything (greedy and self centered), they're not stupid. They probably have had controllers for HBM and HMC ready for a while now just in case.

If not, well, they're screwed.

 

[96Firebird]

People really think AMD holds any sort of control over who SK Hynix can sell HBM to? Am I reading this correctly? D:

I guess we'll have to "Wait and see!" (TM)

 

[Keysplayr]

People really think AMD holds any sort of control over who SK Hynix can sell HBM to? Am I reading this correctly? D:

I guess we'll have to "Wait and see!" (TM)

I see what you did there.

 

[Arachnotronic]

People really think AMD holds any sort of control over who SK Hynix can sell HBM to? Am I reading this correctly? D:

I guess we'll have to "Wait and see!" (TM)

Great business move on the part of SK Hynix -- restrict itstotal market to the distant #2 in terms of graphics market share!

I honestly can't believe that anybody thinks HBM/HBM2 is an "AMD-exclusive" creation.

 

[Blitzvogel]

I was at the Fry's in Austin, TX, on Sunday and they were just about bone dry in terms of higher-end AMD stuff, even R9 280s. I assume it's a supply issue thanks to the product family change, but it was surprising never-the-less. I'm not sure if I should be optimistic about the Fury supply issue......

 

[Azix]

Great business move on the part of SK Hynix -- restrict itstotal market to the distant #2 in terms of graphics market share!

I honestly can't believe that anybody thinks HBM/HBM2 is an "AMD-exclusive" creation.

always with this simplistic market share argument. It really does not matter to Hynix. If AMD buys up all their Chip capacity, even if amd was number 500000000, its money in their pockets. This silly market share is less so amd can't do anything sentiment is for the simple minded.

Plus they probably should be thankful AMD created it with them and they are the first to offer it (possibly only to offer for a year or more). It doesnt have to be a matter of exclusivity. All AMD has to have is a contract for hynix chips and that would shut out nvidia if nobody else was making them. Similar things probably happened with actual GPU chips due to TSMC capacity, in the past