Confirmed by AMD & Intel - Rivals Intel and AMD Team Up on PC Chips to Battle NVidia

Page 11 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

FIVR

Diamond Member
Jun 1, 2016
3,753
911
106
Because HBM is a niche solution, not a mainstream one, integrating HBM MC into RR/PR would have just been a waste of die space. HBM really only makes sense for GPUs at this time.

Why not do the same thing they just did with this Intel chip and integrate a Vega or Polaris GPU +HBM with a Ryzen CPU? The same thing that can be done with EMIB can be done with interposer. That would still use less power than the intel solution and it would allow AMD to sell its own CPUs and gain mindshare and marketshare over both AMD and nvidia, instead of just the latter.
 

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
What if AMD & Ryzen already lost this OEM (Apple) bid for the CPU? Recuperating the GPU portion instead of a complete loss is still a great gain.
Naa it was never in the running. Emib seems like a cheaper option to me than a glorified fat vega apu on some not so cheap imposer only a few would buy due to high cost.
This isnt the most elegant solution but to me it looks very space and tco effective vs the alternative. Its plenty expensive as it is even for apple. These fat apu is 7nm products if imposer cost can even get there and is next gen console stuff.
 

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
Why not do the same thing they just did with this Intel chip and integrate a Vega or Polaris GPU +HBM with a Ryzen CPU? The same thing that can be done with EMIB can be done with interposer. That would still use less power than the intel solution and it would allow AMD to sell its own CPUs and gain mindshare and marketshare over both AMD and nvidia, instead of just the latter.

Who says they won't? I wouldn't be the least bit surprised to see these AMD GPUs paired with HBM, show up in laptops, without the attached Intel CPU.

I actually expect it. The "semi-custom" likely means AMD will be reusing the GPU design for other sales.

The main issue is that the interposer is more expensive than EMIB, but AMD should still be able to undercut Intel.
 

raghu78

Diamond Member
Aug 23, 2012
4,093
1,475
136
This above is all conjecture that you are posting as fact.

https://www.anandtech.com/show/9612...ion-radeon-whole-once-more-led-by-raja-koduri

"Finally, in the longer term, what remains to be seen is if Raja can reverse what has without a doubt been a difficult period for AMD. I think it would be remiss to point out that this kind of a reorganization would also be the precursor to splitting AMD and selling/merging parts of it, however from everyone I’ve talked to this doesn’t appear to be the plan. Rather there is a lot of energy over at AMD backing this reorganization, and that by making graphics whole again and putting it under the steady hand of Raja Koduri, AMD believes they can reinvigorate their graphics group. In the end time will tell, and for the time being there is a great deal of hope that in making Radeon whole once more AMD can carry on in what has been a 30 year legacy of graphics hardware development."

There you have it. AMD created RTG to reinvigorate their graphics group and their GPU business.
 

raghu78

Diamond Member
Aug 23, 2012
4,093
1,475
136
Who says they won't? I wouldn't be the least bit surprised to see these AMD GPUs paired with HBM, show up in laptops, without the attached Intel CPU.

I actually expect it. The "semi-custom" likely means AMD will be reusing the GPU design for other sales.

The main issue is that the interposer is more expensive than EMIB, but AMD should still be able to undercut Intel.

Vega 11 is expected to arrive in H1 2018 and will be targetted at notebooks and desktops unlike Vega 10 which was targetted at high end desktops, workstations and servers.
 

FIVR

Diamond Member
Jun 1, 2016
3,753
911
106
Who says they won't? I wouldn't be the least bit surprised to see these AMD GPUs paired with HBM, show up in laptops, without the attached Intel CPU.

I actually expect it. The "semi-custom" likely means AMD will be reusing the GPU design for other sales.

The main issue is that the interposer is more expensive than EMIB, but AMD should still be able to undercut Intel.

I don't see them putting in the effort if they already are selling the same essential product with intel CPUs for Apple (or whomever). Maybe they will, but at that point it would be a waste of time to do so. OEMs will already be set up to sell this intel/AMD product and customers will already be conditioned to see AMD as the lower-cost, lower-performance option vs intel as the high performance one.


I guess it's possible they will in 2019 when they have a 7nm product but part of my hopes for AMD was that they might get Apple to use their CPUs. This release seems to shut the door on that possibility, at least for the time being.
 

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
I don't see them putting in the effort if they already are selling the same essential product with intel CPUs for Apple (or whomever). Maybe they will, but at that point it would be a waste of time to do so. OEMs will already be set up to sell this intel/AMD product and customers will already be conditioned to see AMD as the lower-cost, lower-performance option vs intel as the high performance one.

Geez, that's a defeatist attitude.

It's not a waste, they can sell their own CPUs with their own GPUs.

It would make no sense at all to NOT offer their own GPUs with their own CPUs.
 

IRobot23

Senior member
Jul 3, 2017
601
183
76
This is VEGA or Polaris architecture?
Anandtech says it is probably Polaris, but Polaris with HBM?
 

neblogai

Member
Oct 29, 2017
144
49
101
I don't see them putting in the effort if they already are selling the same essential product with intel CPUs for Apple (or whomever). Maybe they will, but at that point it would be a waste of time to do so. OEMs will already be set up to sell this intel/AMD product and customers will already be conditioned to see AMD as the lower-cost, lower-performance option vs intel as the high performance one.

You were saying AMD CPUs are better, and should be used instead of Intel. You are also saying AMD will be seen as lower cost, lower-performance vs Intel because of this Intel part. I don't get how you get such conclusion when AMD can put together Ryzen Mobile with Vega11: better CPU (your opinion) with better GPU, and easily beat this.
 
May 11, 2008
19,303
1,129
126
IMHO, this is a great move from AMD + RTG.

Intel sees Nvidia as the real threat because the deep learning and data centres are where the money is.
Nvidia has a huge revenue and can leverage more financial power compared to AMD and as such is a more threatening adversary. Besides the friction between Intel and Nvidia, I think Intel of the last decade has a soft spot for AMD.
Funny thing is that because Intel has higher prices because of the margins, that indirectly thay also kept AMD afloat. If Intel would have sold their cpu's for the same prices as AMD,
everybody who has a smaller budget would also have gone for Intel depriving AMD from selling any cpu.
Intel has a better relationship with AMD, and this is actually a win win situation for AMD / RTG.
because they will sell more gpus. Have Intel + AMD brand recognition. This is very good for AMD / RTG in the long run because it means financial income in the long run.
The laptop market is very big, so this makes sense.
For AMD this is another custom design but if it is any gcn or ncu alike, it also could mean more HSA adoption and more use of GCN gpu derivate as a general compute source.
Nvidia has been very smart with promoting cuda as they have done. Now cuda is a household name in the pc programming world for various disciplines.
And AMD could use the same effect very much.

There is also another thing, This might result and i hope it does that Intel will license EMIB technology to AMD and that AMD will make active use for it.
Because compared to the interposer technology, this may have some benefits. Because silicon interposers are fragile, and there is a asic(gpu or memory or cpu) to silicon interposer connection.
And another connection between the silicon interposer and the package substrate. These are two fail mechanisms that have to be solved by using solutions such as underfill but it all is very critical.
For as far as i understand, Intels emib techonology might be more robust.
Nvidia is researching glass, organic and silicon interposer technology to solve various hurdles they encounter with this technology.

But there is also something that i have been wondering about. can a silicon interposer contain active circuitry or is it just a silicon pcb ?
I am not sure if Intel has stated this, but i would expect the silicon bridge in the emib technology in the near future to contain active circuitry like for example
high speed crossswitches or serdes to link several dies together.

I think that maybe AMD for their future epyc lines might go for emib and maybe use the silicon bridge as an Infinity fabric sub station /cross switch to connect several dies together or
something like that or to peripherals . Or the bridge can be used to solve protocol differences between older and newer dies that can be intermixed.
For the server market and deep learning where the big financial rewards are present this might be good stuff and in the end will also be standard issue for ever more powerfull laptops and end up in what will then exist as desktops.

I predict to not sell your shares, you will very likely regret it.


edit:
Removed typing errors.
 
Last edited:

FIVR

Diamond Member
Jun 1, 2016
3,753
911
106
You were saying AMD CPUs are better, and should be used instead of Intel. You are also saying AMD will be seen as lower cost, lower-performance vs Intel because of this Intel part. I don't get how you get such conclusion when AMD can put together Ryzen Mobile with Vega11: better CPU (your opinion) with better GPU, and easily beat this.

I said they use less power and perform comparably to intel CPUs. Yes, they will be seen as the "lower cost" option because the laptops you will see with Raven Ridge will be much less expensive, much lower performance and probably lower volume than whatever they produce with intel for Apple.

If the only way AMD can get into Apple computers is by being tied to an intel CPU, then Ryzen will never be used by Apple and it will be viewed by Microsoft and others as a low-cost alternative to intel only to be used in cheap, low-margin products. This will hurt consumer perception of AMD Ryzen and will not help AMD in the long run in taking marketshare from intel.

You cannot take marketshare from intel by selling intel CPUs. I would think that is a simple concept to understand. Perhaps others here believe AMD's main competitor is nvidia and not intel, I disagree.
 

LightningZ71

Golden Member
Mar 10, 2017
1,627
1,898
136
From what I'm seeing around, it seems to be a lightly modified Polaris chip.

As for why didn't AMD do This? Cost and demand. This is not going to be a cheap product to manufacture. For AMD, it would be a package near to TR in size and complexity at the present moment. Given both time and a solid revenue stream, I Don't see where AMD would be excluded from developing a an interposer with an 8-12 core 7nm die, possibly with a small iGPU for low power situations, a vega die, and an HBM or 3 stack. By then, the market will have a better understanding of those sorts of products and more OEMs will be willing to take the risk there. I just Don't see anyone being willing to take on such a risk with AMD at this moment. This Intel venture is just what the market needs, a proof of concept.

Why hasn't AMD made a RR die on an interposed with HBM? Cost again. The GPU on RR is sized for its expected available RAM bandwidth and use cases. If you suddenly threw 8 times it's current memory bandwidth at it in its current form, you wouldn't even get a 50% performance boost unless in the most theoretical, memory bound situations. To take advantage of the same stack, the iGPU section on RR would have to be 4 times it's current size. That would be a MUCH larger die which would tank yields, skyrocket per die costs, and be wholly inappropriate for the volume market that RR is currently aimed at. Remember, until 6 months ago, AMD was suffering on the CPU side from revenue stagnation. It doesn't have the money to fund a giant die like that for such a low volume use case.

7nm can make a lot of difference for them. It roughly doubles the floor space for Zepplin at similar chip dimensions. This can enable a 2 CCX, double size GCX and expanded caches in the same space as the RR die (roughly). Couple that with faster RAM and you've got something. Or, you can reduce the die size by 33%, keep a single CCX, double the gcx, mount it on an interposer with an HBM stack and be near the performance of the discussed Intel solution for less money in the same AM4 footprint.

Something else to consider, an AND interposer solution would likely use infinity fabric to link the GPU and gpu. That brings a lot more potential bandwidth between the GPU and external GPU, meaning that there can be flexibility with memory pooling. Using AMD's research in HMA, they could dynamically use all of the memory as system RAM or vram as needed with lower latency and bandwidth penalties as compared to PCIe solutions.
 
  • Like
Reactions: USER8000

neblogai

Member
Oct 29, 2017
144
49
101
If the only way AMD can get into Apple computers is by being tied to an intel CPU, then Ryzen will never be used by Apple and it will be viewed by Microsoft and others as a low-cost alternative to intel only to be used in cheap, low-margin products. This will hurt consumer perception of AMD Ryzen and will not help AMD in the long run in taking marketshare from intel.

You are assuming Ryzen Mobile is more power efficient than Intel. And that might be true for 100% load at some TDPs- but how would it help AMD's prestige and public perception if next Apple's laptop (now all AMD) got 25% shorter battery life at low load tasks- playing videos and browsing? Which is to be expected from Raven Ridge. AMD certainly knows their capabilities and it is probably wise to compete where Ryzen Mobile and Vega are strong, not where they are weak and not ready.
 
  • Like
Reactions: prtskg

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
If this was a product just for Apple - woudn't it be Apple making the announcement, not the Intel/AMD? To me, this looks like a direct successor (and upgrade) of Intel's high power products with Iris graphics, and sold to everyone. Also, Intel said they will be managing the drivers provided to them by AMD- which would not be needed if this was used just by Apple.

Intel has to "manage the drivers" because the iGPU is still present. They have to handle switching between the two GPUs. Nobody else is going to be in the market for this graphics solution mostly due to cost. Don't expect it outside of Apple products.

I don't see why Apple would make the announcement, since what they are getting is basically the same thing they've had recently - Intel CPUs with AMD GPUs. Just in a different package.

You cannot take marketshare from intel by selling intel CPUs.

1). AMD already "sells Intel CPUs" in existing Apple products featuring AMD GPUs. They have done so for years. Nothing changes here.
2). AMD probably can't produce this product, at all. EMIB brought the price down low enough to make it work. For whatever reason, AMD's available solutions may not be able to operate in that cost/power envelope.
3). Apple has been using Intel - not AMD - CPUs in their laptop/desktop products for awhile. I see no indicator that Apple is interested in stopping this relationship in favor of AMD CPUs.

If you think Apple is interested in nixing one x86 CPU supplier in favor of another, think again. Apple is likely to adopt their own CPU tech sooner or later. They may also dump AMD somewhere around that time.
 

maddie

Diamond Member
Jul 18, 2010
4,723
4,628
136
IMHO, this is a great move from AMD + RTG.
..............................................................................................
But there is also something that i have been wondering about. can a silicon interposer contain active circuitry or is it just a silicon pcb ?
I am not sure if Intel has stated this, but i would expect the silicon bridge in the emib technology in the near future to contain active circuitry like for example
high speed crossswitches or serdes to link several dies together.
Yes, there can exist an active interposer.

I really think the hype on EMIB is a little too extreme, because EMIB can't transfer signals directly between 2 hop ICs means that the signal routes in the intermediate IC must be custom designed for that conglomeration of ICs, ie, custom masks for the base metal layers. This, as I see it, reduces general reuseability. A full interposer allows the sub-unit ICs to be reused for many designs with only the interposer routing changing to suit the mix.

Its far cheaper to make a custom interposer than design & fab custom ICs. As I have posted, SI costs are around $1 per 100mm^2.
 

itsmydamnation

Platinum Member
Feb 6, 2011
2,744
3,077
136
You are assuming Ryzen Mobile is more power efficient than Intel. And that might be true for 100% load at some TDPs- but how would it help AMD's prestige and public perception if next Apple's laptop (now all AMD) got 25% shorter battery life at low load tasks- playing videos and browsing? Which is to be expected from Raven Ridge. AMD certainly knows their capabilities and it is probably wise to compete where Ryzen Mobile and Vega are strong, not where they are weak and not ready.
Why is that to be expected? I would call that unexpected. RR has a newer video decode engine then Vega with VP9 support so that is one major issue addressed. im honestly scratching my head at anything else?
 

maddie

Diamond Member
Jul 18, 2010
4,723
4,628
136
I said they use less power and perform comparably to intel CPUs. Yes, they will be seen as the "lower cost" option because the laptops you will see with Raven Ridge will be much less expensive, much lower performance and probably lower volume than whatever they produce with intel for Apple.

If the only way AMD can get into Apple computers is by being tied to an intel CPU, then Ryzen will never be used by Apple and it will be viewed by Microsoft and others as a low-cost alternative to intel only to be used in cheap, low-margin products. This will hurt consumer perception of AMD Ryzen and will not help AMD in the long run in taking marketshare from intel.

You cannot take marketshare from intel by selling intel CPUs. I would think that is a simple concept to understand. Perhaps others here believe AMD's main competitor is nvidia and not intel, I disagree.
You are arguing that AMD should battle both Titans [Intel & Nvidia] at once. A rather risky proposition. This approach of incrementally winning volumes and revenue actually seems to have the best cost/benefit ratio, which is what really wins wars and not hoping for that special knockout punch.
 

neblogai

Member
Oct 29, 2017
144
49
101
Why is that to be expected? I would call that unexpected. RR has a newer video decode engine then Vega with VP9 support so that is one major issue addressed. im honestly scratching my head at anything else?

1) Ryzen Mobile slides show RR having only slightly (~20%?) better efficiency than Bristol Ridge with h264. Efficiency with VP9- lower.
2) Real life laptops with Bristol Ridge and ~50Wh battery are able to play h264 for ~5hours (notebookcheck.net tests), 8550U- for ~8hours.
So even if laptop with Raven Ridge can play h264 for 20% longer- that is about 6h vs 8h battery life, -25% disadvantage. On top of that- Intel was boasting VP9 decoding at 0.4W with their 8th generation CPUs.
 
May 11, 2008
19,303
1,129
126
Yes, there can exist an active interposer.

I really think the hype on EMIB is a little too extreme, because EMIB can't transfer signals directly between 2 hop ICs means that the signal routes in the intermediate IC must be custom designed for that conglomeration of ICs, ie, custom masks for the base metal layers. This, as I see it, reduces general reuseability. A full interposer allows the sub-unit ICs to be reused for many designs with only the interposer routing changing to suit the mix.

Its far cheaper to make a custom interposer than design & fab custom ICs. As I have posted, SI costs are around $1 per 100mm^2.

Oke, but how is it with the total cost and reliability. I have been reading that silicon interposers are fragile for temperature changes and effects like warping must be fully controlled or the microbumps between the asics and interposer or substrate and interposer might come loose. Underfill seems to alleviate that but mounting for example a heatsink is a difficult task. And everything that needs to be done precise and with strict tolerance becomes more expensive.
I understand a silicon interposer itself is cheap, but the whole production process towards the end product may be more expensive. At least that is my understanding of it.

As a sidenote the silicon die bridges in EMIB may not be so bad when thinking of pass through when a designer uses standardized modular technology to connect it all together.
And that is what AMD has done with IF. And when looking at the picture , one only sees 2 chips, but i wonder if there is no limitation to connect 4 dies at once with one square silicon bridge.
Then it may not be that bad as you might think. I do have to note, i do not know the maximum size a silicon bridge in EMIB technology can be.

EDIT

T
he silicon bridge die could perhaps also be a hexagon or octagon if the cpu dies are small and rectangular enough. That would allow for 6 or 8 chips to connect to one silicon bridge die. All at once.
Would be like spokes on a wheel. How many cores does one want ?

wheel-with-spokes.jpg
 
Last edited:

Despoiler

Golden Member
Nov 10, 2007
1,966
770
136
One interesting thing to note. The current Apple Macbook Pro touted AMD's thin packaging technology as necessary to achieve their slimness. Nvidia doesn't have anything like it. They wouldn't have been an option for this initiative with Intel.
 

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136

Well we now have the "source" of the story that AMD reorganized and created RTG to make Intel happy: Crazy rampant speculation from an analyst.

Seriously, its amazing people get paid to make this stuff up:

"McGregor speculated that Intel chose to partner with AMD over Nvidia because AMD has "tried to play nice," even separating the graphics group into a separate entity — Radeon Technology Group — likely to satisfy any Intel concerns."
 
  • Like
Reactions: krumme and raghu78

Despoiler

Golden Member
Nov 10, 2007
1,966
770
136
Well we now have the "source" of the story that AMD reorganized and created RTG to make Intel happy: Crazy rampant speculation from an analyst.

Seriously, its amazing people get paid to make this stuff up:

"McGregor speculated that Intel chose to partner with AMD over Nvidia because AMD has "tried to play nice," even separating the graphics group into a separate entity — Radeon Technology Group — likely to satisfy any Intel concerns."

I lol'd when I read that also. Analysts are the absolute worst.
 
Last edited:
  • Like
Reactions: raghu78 and IEC