AMD Rx 400 series code-name "Arctic-Islands" May only sport HBM 1 and not HBM 2

[Memories4K]

Sources-[Link] [Link]

 

 

 

"The latest rumors for AMD’s next generation graphics card has surfaced from TweakTown. From the looks of it, things aren’t running really great for AMD after they failed to negotiate an equity deal with Silver Lake in 2015 (Summer) and the discussions were put on hold, now it looks like another bad news has emerged as AMD faces a lot of troubles sourcing out next generation HBM2 chips for use in their high-end Arctic Islands GPUs arriving in 2016."

 

 

If this is true it doesn't look good for AMD.

Restricted to 4GB another card generation while Nvidia gets the run on HBM 2.

Let's hope rumor's false and AMD can source HBM 2, not just get the short and lower yields; i could've swore they had first dibs though.

[dizmo]

Haha, well then. AMD really is just getting closer and closer to that iceberg aren't they.

Shame, I was really hoping they'd be able to pull off a fantastic comeback with a new CPU and HBM2.

[Fetzie]

An source I'm not going to disclose told me that Intel was going to give us a 10 core consumer chip in the generation after Cannonlake.

 

There, a rumour with about as much substance as this one.

 

(nobody told me that, I made it up)

 

 

If SK-Hynix has the ability to supply both AMD and nVidia, there won't be a shortage.

[Guest Kloaked]

This is a rumor. Don't take this as canon.

[christianled59]

Damn... I kind of hope nvidia doesn't dominate the next gen gpus.. I hope amd can stay on top in dx12 at least.

[aneil1998]

An source I'm not going to disclose told me that Intel was going to give us a 10 core consumer chip in the generation after Cannonlake.

 

There, a rumour with about as much substance as this one.

 

(nobody told me that, I made it up)

 

 

If SK-Hynix has the ability to supply both AMD and nVidia, there won't be a shortage.

There might be a shortage for AMD if SK-Hynix is producing it because Nvidia GPU's will be in higher demand that AMD's.

[Fetzie]

Damn... I kind of hope nvidia doesn't dominate the next gen gpus.. I hope amd can stay on top in dx12 at least.

It is an unsubstantiated rumour from an "anonymous source". i.e. treat it as bullshit until proven otherwise.

[Memories4K]

 

If SK-Hynix has the ability to supply both AMD and nVidia, there won't be a shortage.

There might be a shortage for AMD if SK-Hynix is producing it because Nvidia GPU's will be in higher demand that AMD's.

AMD's had trouble getting HBM for their Fury lineup before.

AMD needs to meet demand; if the mid-lower range of cards use HBM 1 then i don't think anyone will care.

But promising HBM 2 and not delivering will be a problem.

[sof006]

Hahahaha expected something like this

[MattLantis]

I like the name

[Memories4K]

I like the name

"Arctic Islands"

I'm only holding off on the Rx 300 series just because i'm curious if next gen AMD GPUs are amazingly efficient or not compared to the mini-volcano that was the 290x.

[Fetzie]

AMD's had trouble getting HBM for their Fury lineup before.

AMD needs to meet demand; if the mid-lower range of cards use HBM 1 then i don't think anyone will care.

But promising HBM 2 and not delivering will be a problem.

There are shortages and issues for every new technology though.

[Mahigan]

Well,

 

Let's use Logical deduction.

 

There are two rumors:

1. AMD will get priority over HBM2 shipments
2. AMD is having trouble sourcing HBM2 memory

 

Why are both of these rumors either patently false or spell disaster for both AMD and nVIDIAs next gen cards? To start, while it is true that AMD holds several patents over HBM2, AMD applied for a JEDEC standard which does not allow for discriminatory behavior. Therefore AMD could not receive priority over the shipments of HBM2 memory as this would count as discriminatory behavior.

For the same reason, if AMD is having issues sourcing HBM2 then so is nVIDIA, because discriminatory behavior is not permitted. This would point to an issue with the production of HBM2 and not an issue with AMD having issues because nVIDIA is sourcing all of the HBM2 (not allowed according to the JEDEC standard).

If nVIDIA attempted this, and Samsung/Hynix gave nVIDIA preferential treatment, then AMD could make a patent claim and apply royalties as well as fines onto nVIDIA/Samsung and Hynix in order to recoup costs (permitted by the JEDEC license due to AMDs patents).

 

So we're left with this rumor being false, in which case neither AMD or nVIDIA will have issues sourcing HBM2 or this rumor is true and it would make it hard for either of the two to source HBM2 due to some production issues.

[VagabondWraith]

Ouch. Would not be good for AMD IF the rumors are to be true. nVidia could run away from AMD if this truly presents a problem.

But still too early to say anything definitive yet.

[ivan134]

"Arctic Islands"

I'm only holding off on the Rx 300 series just because i'm curious if next gen AMD GPUs are amazingly efficient or not compared to the mini-volcano that was the 290x.

GCN 1.2 is already amazingly efficient. It can only get better from there with a node shrink.

[Memories4K]

GCN 1.2 is already amazingly efficient. It can only get better from there with a node shrink.

Not Arctic enough to justify the cool/edgy name. (Joking)

I am impressed with the Rx 300 series though.

[Notional]

Well,

 

Let's use Logical deduction.

 

There are two rumors:

1. AMD will get priority over HBM2 shipments
2. AMD is having trouble sourcing HBM2 memory

 

Why are both of these rumors either patently false or spell disaster for both AMD and nVIDIAs next gen cards? To start, while it is true that AMD holds several patents over HBM2, AMD applied for a JEDEC standard which does not allow for discriminatory behavior. Therefore AMD could not receive priority over the shipments of HBM2 memory as this would count as discriminatory behavior.

For the same reason, if AMD is having issues sourcing HBM2 then so is nVIDIA, because discriminatory behavior is not permitted. This would point to an issue with the production of HBM2 and not an issue with AMD having issues because nVIDIA is sourcing all of the HBM2 (not allowed according to the JEDEC standard).

If nVIDIA attempted this, and Samsung/Hynix gave nVIDIA preferential treatment, then AMD could make a patent claim and apply royalties as well as fines onto nVIDIA/Samsung and Hynix in order to recoup costs (permitted by the JEDEC license due to AMDs patents).

 

So we're left with this rumor being false, in which case neither AMD or nVIDIA will have issues sourcing HBM2 or this rumor is true and it would make it hard for either of the two to source HBM2 due to some production issues.

 

While I agree with most of your post, I do believe the "no discriminatory behaviour" is targeted towards manufacturers not being allowed access/licenses for the JEDEC tech, and not in what contracts are being written between vendor and customer, as the latter is common business strategies (priority access). Maybe pricing differences are a no no, but that is not what AMD supposedly has with SK Hynix.

 

 

Either way, this news makes no sense and has nothing to back it up.

 

 

AMD's had trouble getting HBM for their Fury lineup before.

AMD needs to meet demand; if the mid-lower range of cards use HBM 1 then i don't think anyone will care.

But promising HBM 2 and not delivering will be a problem.

 

Cannot be compared as SK Hynix is the only vendor on the market to actually manufacture HBM. So there is only 1 supplier of the tech. Their supply issues are the results of new technology with subsequent lower yields and lower manufacturing capabilities (we already know SK Hynix is increasing manufacturing capabilities for HBM1.

 

I do agree that HBM1 might be used for lower to mid end cards, if it pays off. There would be no issue with that what so ever, as these cards would never max out HBM1 bandwidth anyways. The issue is more whether we will even see 4GB cards next year. Maybe not.

[GlassBomb]

What I were able to make out from what little I read.

 

HBM1 will be used for the mid- to high-end GPU's. Where as the HBM2 will be used for high end/enthusiast. Where you really do need more than 4GB's of vRAM.

[Curufinwe_wins]

Well,

Let's use Logical deduction.

There are two rumors:

• AMD will get priority over HBM2 shipments
• AMD is having trouble sourcing HBM2 memory

Why are both of these rumors either patently false or spell disaster for both AMD and nVIDIAs next gen cards? To start, while it is true that AMD holds several patents over HBM2, AMD applied for a JEDEC standard which does not allow for discriminatory behavior. Therefore AMD could not receive priority over the shipments of HBM2 memory as this would count as discriminatory behavior.

For the same reason, if AMD is having issues sourcing HBM2 then so is nVIDIA, because discriminatory behavior is not permitted. This would point to an issue with the production of HBM2 and not an issue with AMD having issues because nVIDIA is sourcing all of the HBM2 (not allowed according to the JEDEC standard).

If nVIDIA attempted this, and Samsung/Hynix gave nVIDIA preferential treatment, then AMD could make a patent claim and apply royalties as well as fines onto nVIDIA/Samsung and Hynix in order to recoup costs (permitted by the JEDEC license due to AMDs patents).

So we're left with this rumor being false, in which case neither AMD or nVIDIA will have issues sourcing HBM2 or this rumor is true and it would make it hard for either of the two to source HBM2 due to some production issues.

Unless due to high demand hynix wants to charge x, nvidia says sure, amd says Bullshit I'm only paying y. That's not discrimination.

[Mahigan]

Unless due to high demand hynix wants to charge x, nvidia says sure, amd says Bullshit I'm only paying y. That's not discrimination.

 

 

Actually it is... 

 

Prices are determined through the consent of all of the JEDEC members.

[NumLock21]

So AMD cannot get the type of ram they developed?

[Notional]

Actually it is... 

 

Prices are determined through the consent of all of the JEDEC members.

 

Which still makes priority supply possible, although it depends on the definition of product quotas.

[Briggsy]

The whole rumor is based on uneducated assumptions, as others here have pointed out. If Nvidia were able to source hbm2 and only them, there would be a serious breach of agreement between everyone involved.

What is likely, is that AMD will be using HBM 1 for all low and mid range cards. The flagship card will likely have hbm2. I would expect the exact setup for Pascal on Nvidia side as well, as 1TB/s bandwidth is stupid on current low end and midrange cards. In fact, entry cards will probably jump from gddr3 to gddr5.

Just when Wccftech was putting out some reliable guesswork, they had to go and push this turd of a rumor out.

[themaniac]

So AMD cannot get the type of ram they developed?

yeah, if they developed it then they should get it

[Guest Kloaked]

So AMD cannot get the type of ram they developed?

 

 

yeah, if they developed it then they should get it

 

It's not that simple.