Samsung and Global Foundries will produce AMD's Next Gen Greenland GPU and Zen CPU

[Arachnotronic]

Do you have a link to proof of this? I'm not even sure Samsung has LPP at all otherwise AMD would be using them earlier than GlobalFoundries instead of later, wouldn't they?

Sorry, I don't have a link. However this is something I was very curious about so I asked friends within the industry who would know. They have all said that it is LPP.

 

[Andrei.]

That's not what several independent industry sources have told me.

As the story goes, Samsung basically used A9 as its yield learning vehicle for 14LPP (Exynos 7420 was used for LPE).

Not what I know of so I disagree. 8890 and 820 should be the first LPP parts.

 

[Arachnotronic]

Not what I know of so I disagree. 8890 and 820 should be the first LPP parts.

I suppose we will agree to disagree 🙂

 

[el etro]

Thevenin from SA forums says 14LPP was going to be released a few months before Iphone 6s launched. Not enough time to bake A9 on LPP process. From what i know, 16FF from TSMC is still not the plus version.

 

[Abwx]

I suppose we will agree to disagree 🙂

We once desagreed when i told you that AMD would have no trouble fabbing samples at Samsung in the waiting of GF being ready.

I also told you that Apple s does not use LPP since it s too leaky, with the possibility that if it s a LPP variant then it s a high threshold voltage variant wich is not representative at all of the exact process that will be used by AMD.

So If i were you i would be more cautious before keeping doing such random statements..

 

[Arachnotronic]

Thevenin from SA forums says 14LPP was going to be released a few months before Iphone 6s launched. Not enough time to bake A9 on LPP process. From what i know, 16FF from TSMC is still not the plus version.

A9 TSMC version is 100% 16FF+; 16FF (non Plus) never went into production.

Anyway, I stand by my statements and believe my sources on the matter to be good, but if you're hearing different from sources you trust, then that's cool too.

 

[Arachnotronic]

We once desagreed when i told you that AMD would have no trouble fabbing samples at Samsung in the waiting of GF being ready.

I also told you that Apple s does not use LPP since it s too leaky, with the possibility that if it s a LPP variant then it s a high threshold voltage variant wich is not representative at all of the exact process that will be used by AMD.

So If i were you i would be more cautious before keeping doing such random statements..

Ah, 14nm LPP is too leaky for A9 but it's all good for Exynos 8890 and Snapdragon 820? OK then 🙂

 

[Abwx]

Ah, 14nm LPP is too leaky for A9 but it's all good for Exynos 8890 and Snapdragon 820? OK then 🙂

If you insist in forgetting what i , uselessly apparently, repeated ad nauseam you ll be left asking unrelentlessly the same question, at some point you should do your homework..

As i told you, there s LPP, LPP, LPP and LPP...

http://www.anandtech.com/show/9764/...e-first-14nm-finfet-sample-production-success

It's LPE.

Actualy it does not even matter, set apart,of course, for some people who are using Apple A9 results as a hope that Samsung LPP is rubbish and that as a consequence AMD s Zen wont entail in Intel bread and butter...

Those who have some logic have long deducted that LPP advantage over LPE is a paltry 15% better perf/watt, wich means that the transistors (trans)conductance has been increased by the same amount.

What matters actualy is that the second grade speed wise of LPP has already roughly 20% better perf/Watt (at 2.4GHz) than Intel s fastest 14nm and that the faster variant should increase this advantage.

 

[Arachnotronic]

Nevermind.

 

[The Stilt]

AFAIK Samsung started offering 14nm LPP in August, for mass production. Previously all the stuff is supposed to be made on LPE. Even GlobalFoundries had the LPE available in Q1.

Would love to know what´s the actual difference between the two (other than the 15% advantage quote from Samsung itself).

 

[Arachnotronic]

Those who have some logic have long deducted that LPP advantage over LPE is a paltry 15% better perf/watt, wich means that the transistors (trans)conductance has been increased by the same amount.

What matters actualy is that the second grade speed wise of LPP has already roughly 20% better perf/Watt (at 2.4GHz) than Intel s fastest 14nm and that the faster variant should increase this advantage.

100% unsubstantiated claim.

 

[el etro]

Would love to know what´s the actual difference between the two (other than the 15% advantage quote from Samsung itself).

They say 14LPP have even smaller FrontEnd/BackEnd. Is a smaller transition, like from 20LPE to the cancelled(due to the imminent readiness of 14FF) 20LPM. See first table here: http://www.anandtech.com/show/9330/exynos-7420-deep-dive

 

[prtskg]

It's odd that some people say A9 was made using lpp and finfet+ process when neither of the two achieved volume production then. TSMC finfet reached volume production in Q2 2015 and they said that finfet+ would reach volume production in Q3 (not that it did).
http://www.taipeitimes.com/News/biz/archives/2015/08/11/2003625049
Samsungs lpp was expected to reach volume production in H2 2015. I'd repeat that Samsung already considers lpp to be important only in 2016 and after.
http://www.samsung.com/semiconductor/foundry/process-technology/14nm/?CID=AFL-hq-mul-0813-11000170

 

[prtskg]

double post. Is there no delete option?

 

[Lepton87]

double post. Is there no delete option?

Sadly As far as I know there's not such an option. Back to the topic. I think that if Apple A9 was produced on a better node that the competing Exynos they wouldn't miss a chance to boast about it.

 

[lopri]

That's not what several independent industry sources have told me.

As the story goes, Samsung basically used A9 as its yield learning vehicle for 14LPP (Exynos 7420 was used for LPE).

Well, from a report I read earlier this year said the 7420's yields were over 50% at the end of last year and 70% during the first quarter 2015. If anything the 7420 was a yield learning vehicle for 14 LPE (and whatever else).

From the links prtskg provided above:

14LPP (Performance boosted edition) is the 2nd FinFET generation which the performance is enhanced up to 10%. 14LPP is the single platform for every application designs with the improved performance for computing/Network designs and the lowered power consumption for Mobile/Consumer designs. 14LPP will be the main process technology offering in 2016 and after.

The underlined part is hard to square with the claim that A9 is produced on 14 LPP. I would presume the first parties have a correct knowledge on such matters and I see no reason as to why Samsung would lie about it. (I won't bother with the claim that the A9 is Samsung's vehicle for learning 14 LPP)

 

[Arachnotronic]

Well, from a report I read earlier this year said the 7420's yields were over 50% at the end of last year and 70% during the first quarter 2015. If anything the 7420 was a yield learning vehicle for 14 LPE (and whatever else).

From the links prtskg provided above:



The underlined part is hard to square with the claim that A9 is produced on 14 LPP. I would presume the first parties have a correct knowledge on such matters and I see no reason as to why Samsung would lie about it. (I won't bother with the claim that the A9 is Samsung's vehicle for learning 14 LPP)

Well, I have confidence in my sources but you are of course free to believe whatever you'd like. I have said what I know and nothing you have posted really impacts what I believe to be true 🙂

I will say though that we are talking about Apple and its flagship iPhone that will likely generate >$100 billion in revenue over its lifetime. Who do you think pushed Samsung and TSMC to develop 14 LPP and 16FF+, respectively? There is no foundry customer that has as much influence on the foundry road-maps as Apple does and there is no (mobile) customer that is more demanding in terms of performance requirements 🙂

 

[krumme]

Amen

 

[coercitiv]

I will say though that we are talking about Apple and its flagship iPhone that will likely generate >$100 billion in revenue over its lifetime. Who do you think pushed Samsung and TSMC to develop 14 LPP and 16FF+, respectively?

The same iPhone that was supposed to get a sapphire screen? Not everything Apple wants turns into product particles, so let's not use that as an argument.

I'd rather trust your sources than believe that Apple can push forward the semiconductor industry through sheer iPhone financial might.

 

[Abwx]

100% unsubstantiated claim.

Lol, I m 100% sure based on my knowledge of transistors physics, and if you do not agree ask someone at Intel, preferably an enginer of course, for the rest you can keep on posting denials...

Sorry, I don't have a link. However this is something I was very curious about so I asked friends within the industry who would know. They have all said that it is LPP.

What is the actual difference between LPE and LPP..?.

I m sure you would be hard pressed to point what it really is about, yet you re making statements as if you had a clear idea.

 

[ShintaiDK]

Based on _your_ knowledge? Sorry abwx, your credentials are 0. That's why people tend to ask you for facts...that you cant deliver.

Hints that they are using LPP as well as TSMC is using 16FF+:
http://www.fool.com/investing/gener...may-have-made-a-huge-mistake-in-having-s.aspx
https://benchlife.info/problem-might-not-from-samsung-a9-soc-maybe-is-apple-issue-10242015/
http://www.nasdaq.com/article/3-imp...bring-to-the-a10-processor-next-year-cm543646

And with Apple ditching LPP in favour of 16FF+. We all know what is the best one. AMD keeps betting on the wrong horse. And the result can be outright disasterous in the GPU section. The CPU section will continue with AMDs inferior nodes and uarchs

 

[Andrei.]

Based on _your_ knowledge? Sorry abwx, your credentials are 0. That's why people tend to ask you for facts...that you cant deliver.

Hints that they are using LPP as well as TSMC is using 16FF+:
http://www.fool.com/investing/gener...may-have-made-a-huge-mistake-in-having-s.aspx
https://benchlife.info/problem-might-not-from-samsung-a9-soc-maybe-is-apple-issue-10242015/
http://www.nasdaq.com/article/3-imp...bring-to-the-a10-processor-next-year-cm543646

Two of those links are by the same author who just speculates and the third one actually quotes Chipworks as it being LPE and 16FF+.

 

[ShintaiDK]

I haven't seen anything yet confirming that A9 uses LPE. So could you supply this? Since its so easy just to refuse anything else. I would like to see an official Apple or Samsung statement.

Considering how Samsungs own 2016 Exynos lineup only contains 1 14nm product and 2 28nm products. I think its safe to say their process is anything but ground breaking.

 

[Andrei.]

You won't get an official statement from either of them.

Considering how Samsungs own 2016 Exynos lineup only contains 1 14nm product and 2 28nm products. I think its safe to say their process is anything but ground breaking.

They have far more SoCs that they don't talk about. There's also other vendors using 14nm. In any case based on info here it's 2:0 in favor of LPE and that LPP is indeed at 2016 node. 16FF+ will still be slightly better than 14LPP but the differences will be smaller it'll not be worthy of argument.

 

[ShintaiDK]

You won't get an official statement from either of them.
They have far more SoCs that they don't talk about. There's also other vendors using 14nm. In any case based on info here it's 2:0 in favor of LPE and that LPP is indeed at 2016 node. 16FF+ will still be slightly better than 14LPP but the differences will be smaller it'll not be worthy of argument.

Why dont you prove it besides just defending Samsung with no end?