Samsung to Fab AMD "Zen" and "Arctic Islands" on its 14 nm FinFET Node

[btarunr]

It has been confirmed that Samsung will be AMD's foundry partner for its next generation GPUs. It has been reported that AMD's upcoming "Arctic Islands" family of GPUs could be built on the 14 nanometer FinFET LPP (low-power Plus) process. AMD's rival NVIDIA, meanwhile, is building its next-gen "Pascal" GPU family on 16 nanometer FinFET node, likely at its traditional foundry partner TSMC.

It gets better - not only will Samsung manufacture AMD's next-gen GPUs, but also its upcoming "Zen" family of CPUs, at least a portion of it. AMD is looking to distribute manufacturing loads between two foundries, Samsung and GlobalFoundries, perhaps to ensure that foundry-level teething trouble doesn't throw its product launch cycle off the rails. One of the most talked about "Arctic Islands" GPUs is codenamed "Greenland," likely a successor to "Fiji." Sales of some of the first chips - GPUs or CPUs - made at Samsung, will begin some time in Q3 2016. Some of the other clients for Samsung's 14 nm FinFET node are Apple and Qualcomm. The company plans to speed up development of its more advanced 10 nm node to some time in 2017.



View at TechPowerUp Main Site

 

[GhostRyder]

So both huh, well it should be interesting at least.

 

[Zakin]

This is actually making it pretty interesting, to me at least.

 

[cdawall]

"We are still pissed at global foundries"

 

[suraswami]

So a Transition for AMD into Samsung hands, hmm by 2017 will it be Samsung vs Intel and Samsung vs NV?

 

[FordGT90Concept]

Sucks for Global Foundries but my guess is Global Foundries is going to keep AMD's console processor production.

AMD will likely even have a process advantage against NVIDIA. Things are really looking up for AMD right now. I hope they deliver.

Samsung vs NV?

Already some of that happening in the ARM market.

 

[renz496]

Sucks for Global Foundries but my guess is Global Foundries is going to keep AMD's console processor production.

AMD will likely even have a process advantage against NVIDIA. Things are really looking up for AMD right now. I hope they deliver.


Already some of that happening in the ARM market.

was samsung 14nm really have big advantage compared to TSMC 16nm FF? plus i will only going to believe that samsung going to make AMD gpu when the actual product hit the market. they say tonga will be made by GF but in the end it did not happen. then the talks starts again when AMD was about to release Fury and yet TSMC still the one end up making it.

 

[RejZoR]

What I'm wondering more is why AMD ditched their mobile GPU division!? Almost every phone has Adreno GPU now. Wouldn't that be like a good market for AMD if they kept it?

 

[renz496]

that depends. adreno has good position in SoC market right now (especially in phones) because of Qualcomm baseband/modem domination in mobile. the way i heard about it qualcomm was forcing phone maker to use their snapdragon package if they want to use qualcomm modem/baseband. samsung exynos used to have qualcomm modem inside them. but the much recent exynos are using intel modem instead. most likely because qualcomm only offering their snapdragon package if you want to use their latest modem.

 

[suraswami]

What I'm wondering more is why AMD ditched their mobile GPU division!? Almost every phone has Adreno GPU now. Wouldn't that be like a good market for AMD if they kept it?

Blame AMD management Idiots!!

 

[Chaitanya]

Blame AMD management Idiots!!

thats the sad that AMD has been in for last few years.

 

[midnightoil]

was samsung 14nm really have big advantage compared to TSMC 16nm FF? plus i will only going to believe that samsung going to make AMD gpu when the actual product hit the market. they say tonga will be made by GF but in the end it did not happen. then the talks starts again when AMD was about to release Fury and yet TSMC still the one end up making it.

Difficult to tell, especially as Samsung (and GF's copy not quite exact) 14nm FF LPP process is a low power process and TSMC's 16nm FF+ is a high power process. But we do know that Samsung's LPE and LPP had vastly lower capacitance and leakage than Intel's (physically smaller) 14nm FF versions used in their mobile chips, also most probably much, much higher yields and much lower cost. I'd guess the same is also true of Samsung vs TSMC, but who knows.

Capacity won't be even remotely comparable. TSMC are very constrained for 16nm FF+. If AMD can get enough HBM2 and interposers, supply of Arctic Islands should be great (Zen too) ... Pascal may be little more than a dribble.

I was hoping AMD would switch to Samsung as their main foundry partner moving forward, and GF secondary, but didn't trust them to do so. My confidence in the new management has improved significantly now.

 

[newtekie1]

AMD is looking to distribute manufacturing loads between two foundries, Samsung and GlobalFoundries

I wonder if we are going to have another chipgate on our hands...

 

[Ferrum Master]

I liked the time when it mattered to pick up gpu silicon either from UMC or TSMC...

 

[midnightoil]

I liked the time when it mattered to pick up gpu silicon either from UMC or TSMC...

UMC produce the interposers and mount the chips on them for Fiji. They will do so for NVIDIA with Pascal, too. Possibility AMD may switch to Samsung for Arctic Islands though.

 

[Blue-Knight]

Samsung, you've just failed one more time.

 

[Slizzo]

Good to hear, means that both GPU makers have a fab that can focus on just their parts (for GPU that is, of course they're fabbing lots of other stuff for other vendors).

Interesting that GloFo and Samsung are going to be making Zen... Wonder if there will be any difference between the two chips, or if they're simply going to be fabbing some of Zen at Samsung, and some at GloFo (talking about parts of the product stack at each, not full stack at both).

 

[Frick]

Samsung, you've just failed one more time.

Why?

 

[truth teller]

hmm, interdasting. samsung devices with amd hardware when?

but seriously, a closer partnership between the two companies (faster manufacturing traded for/with hardware design) would be sweet for both sides of the deal

 

[HumanSmoke]

I liked the time when it mattered to pick up gpu silicon either from UMC or TSMC...

Back then UMC still had a seat at the feast (from memory their 80nm was used by ATI/AMD, and their 65nm/55nm was interchangeable with TSMC's - a lot of Nvidia G200/G92's came out of UMC). I think they spent too much time dithering over whether to license Samsung's process or develop their own - which they ended up doing - but like SMIC, the pace of development has rather left them at the children's table.

Interesting that GloFo and Samsung are going to be making Zen... Wonder if there will be any difference between the two chips, or if they're simply going to be fabbing some of Zen at Samsung, and some at GloFo (talking about parts of the product stack at each, not full stack at both).

From what I've read, GloFo's 14nmLPP process is identical to the Samsung process that forms the basis for the license. On the surface this makes sense. Tweaking a process node (whether interconnect pitch, gate pitch/length) requires extra R&D, test/validation, and most importantly, time to market. The intended customer(s) would have to contract for considerable wafer volume to offset costs -and be willing to accept delays over the vanilla process- to make the scenario worthwhile.

 

[geon2k2]

What I'm wondering more is why AMD ditched their mobile GPU division!? Almost every phone has Adreno GPU now. Wouldn't that be like a good market for AMD if they kept it?

Actually Qualcomm bought Adreno tech from AMD and you can see Adreno is just an anagram of Radeon. (same letters arranged differently)

 

[Dave65]

WOW, I didn't see that coming.

 

[TheLostSwede]

I guess what most of those of you commenting here has missed is the fact that the chip design platform is shared between GloFo, Samsung, TSMC and IBM. This was done to make it easier for their customers to switch between the four manufacturers if needed. Now I don't know exactly how easy it is, but this was done a few years ago to make sure that the companies could help fill gaps that some of them might have. This is really not surprising news if you've been following things for a few years. I'm sure I can dig up some old slides from GloFo about this if I have to.

 

[HumanSmoke]

I guess what most of those of you commenting here has missed is the fact that the chip design platform is shared between GloFo, Samsung, TSMC and IBM. This was done to make it easier for their customers to switch between the four manufacturers if needed.

That should be readily apparent I would have thought given that Apple's A9 can be easily ported between TSMC's CLN16FF and Samsung's 14nmLPE. Presumably the same design rules apply for 16nmFF+/FFC and 14nmLPP.

 

[c2DDragon]

Q3 2016 ??? Ouch ! I can't wait to see the Green VS Red 2016 fight.