AMD's Next Gen Products Taped Out, Possibly Delayed

[childofthekorn]

Pascal in the URL? Attempt at click bait?

 

[Mousemonkey]

Another thing mentioned in the article that is misreported is possible technical difficulties with the process node .
Its actually more mature than intels 14 nm . Samsung are licensing the tech to Global Foundries , and Samsung have been at 14 nm longer than Intel have .

Its very hard to find something that is factual in the article but I will presume the authors name is correct

There is just so much wrong here.

Samsungs first 14nm chip was tapped out in 2012 and it was just a test chip. Their first actual chip was not released until 2015 with the Galaxy S6 line.

Intel had 14nm tapped out in 2011 and started building their FAB the same year, FAB 42 in Chandler. Intels first chips with 14nm was demoed in 2013, a Ultrabook with an actual working product not a test chip, and was in full shipping production in 2014.

To top it off, Samsungs 14nm is not geared towards high power chips, that has to be tweaked by GloFlo for that purpose while Intels 14nm is geared towards high power chips. Yields will easily be better for Samsung with low power, more simplistic CPUs vs a higher power and more complex CPU. AMD will not get the same yields out of Samsungs 14nm as Samsung does.

Dude, you're dealing with AMD fanatics so why would you expect anything less? :lol:

 

[TechyInAZ]

The X86 license is non-transferable should AMD be purchased.

Speaking of that, I still wonder what's going on with Microsoft buying AMD? I heard they were thinking about that a few months ago.

 

[gamerk316]

The X86 license is non-transferable should AMD be purchased.

Speaking of that, I still wonder what's going on with Microsoft buying AMD? I heard they were thinking about that a few months ago.

As I said then, given MSFT wants nothing to do with hardware anymore, it's not happening. The roumer was started mainly by tech sites as a "well, MSFT could buy AMD, it makes sense", and people started to report it as fact.

 

[palladin9479]

The X86 license is non-transferable should AMD be purchased.

This has already been debunked. The x86 License is 100% secure as long as AMD remains incorporated independently. It doesn't matter who owns AMD nor who's sitting on the board of directors or who it's CEO is, as long as they are a separate economic entity then all property, business agreements, and licensing agreements remain intact and valid. Which make sense if you stop and remember that AMD's owners are changing very day as stock is bought or sold. Should a large entity come in and buy up over 50% total stock they would gain effective control over AMD yet the company would still remaining a separate entity. They could fold AMD into a holdings company yet it would still remain separate. It's only when it's unincorporated that things change, which is what AMD did to ATI.

For example, if Samsung bought AMD from it's current owners, meaning they bought up at least 51% of the stock, Samsung would then own and control AMD. Samsung then could have AMD produce x86 CPU's all day every day, Samsung could funnel funds into AMD to do this, Samsung could send over Engineers or transfer technology into AMD, could establish unilateral cross licensing agreements and so forth. But in no circumstances could Samsung take AMD's x86 license and design / manufacture x86 CPU's themselves.

 

[alextheblue]

I think what annoyed me enough to say something is the thought of what impact a headline like yours has - especially at such a large and influential site like Tom's. I mean look at the comment section. Most people just assumed that it was delayed. "Time to short AMD again" - we're even talking financial influence here. There wasn't even a Rumor, Op-Ed, Speculation, or Mad Ravings tag on the title to warn folks. Half of the people don't dig far enough into a story to tell what's fact and what's uh... let's be diplomatic here... rumor! Yes that's it, rumor.

....except that our EIC interviewed AMD's CEO, Lisa Su, and was told that Zen would come out sometime in 2016. If it doesn't come out until Q1 2017 then it is still delayed.

http://www.tomshardware.com/news/amd-ceo-lisa-su-interview,29327.html

I said that "even very early 2017 would be roughly on schedule" for this type of product. Roughly, approximately, not far off the mark. Now, kindly point to the evidence you've uncovered of Zen being delayed. Or else tag it with Rumor. Or write an article about Pascal "Possible Delays" based on the same kind of hard hitting evidence you've used to fuel your Zen delay rumor. Why not, it'll get clicks.

I know you're not going to cede anything regardless of the actual facts (or lack thereof). It got clicks, that's all you care about. But I don't think I'm the only one who is going to take any of your future uh, "news" articles with salt from here on in.

 

[IInuyasha74]

I think what annoyed me enough to say something is the thought of what impact a headline like yours has - especially at such a large and influential site like Tom's. I mean look at the comment section. Most people just assumed that it was delayed. "Time to short AMD again" - we're even talking financial influence here. There wasn't even a Rumor, Op-Ed, Speculation, or Mad Ravings tag on the title to warn folks. Half of the people don't dig far enough into a story to tell what's fact and what's uh... let's be diplomatic here... rumor! Yes that's it, rumor.

....except that our EIC interviewed AMD's CEO, Lisa Su, and was told that Zen would come out sometime in 2016. If it doesn't come out until Q1 2017 then it is still delayed.

http://www.tomshardware.com/news/amd-ceo-lisa-su-interview,29327.html

I said that "even very early 2017 would be roughly on schedule" for this type of product. Roughly, approximately, not far off the mark. Now, kindly point to the evidence you've uncovered of Zen being delayed. Or else tag it with Rumor. Or write an article about Pascal "Possible Delays" based on the same kind of hard hitting evidence you've used to fuel your Zen delay rumor. Why not, it'll get clicks.

I know you're not going to cede anything regardless of the actual facts (or lack thereof). It got clicks, that's all you care about. But I don't think I'm the only one who is going to take any of your future uh, "news" articles with salt from here on in.

Dude what facts are you arguing? I get it, you love AMD. Stop ranting on because there was some news about AMD which wasn't positive. It isn't even all that negative. AMD's CEO stated that it would be out in 2016 a few months ago, and now the company's CFO said that it will be out most likely 12-14 months from when it was taped out, which was just in the last few months. Q1 2017 is just a little over 12 months away, and if it is released in Q1 2017, then that means it was delayed. Are you really incapable of understanding simple concepts like time, in order to know that that 2017 is after 2016?

The information that I have to support this is listed in the article and in my comments, it comes from a credible source, and there is nothing about it which can be construed as a rumor in any way.

 

[Outlander_04]

Another thing mentioned in the article that is misreported is possible technical difficulties with the process node .
Its actually more mature than intels 14 nm . Samsung are licensing the tech to Global Foundries , and Samsung have been at 14 nm longer than Intel have .

Its very hard to find something that is factual in the article but I will presume the authors name is correct

There is just so much wrong here.

Samsungs first 14nm chip was tapped out in 2012 and it was just a test chip. Their first actual chip was not released until 2015 with the Galaxy S6 line.

Intel had 14nm tapped out in 2011 and started building their FAB the same year, FAB 42 in Chandler. Intels first chips with 14nm was demoed in 2013, a Ultrabook with an actual working product not a test chip, and was in full shipping production in 2014.

To top it off, Samsungs 14nm is not geared towards high power chips, that has to be tweaked by GloFlo for that purpose while Intels 14nm is geared towards high power chips. Yields will easily be better for Samsung with low power, more simplistic CPUs vs a higher power and more complex CPU. AMD will not get the same yields out of Samsungs 14nm as Samsung does.

Dude, you're dealing with AMD fanatics so why would you expect anything less? :lol:

The point of course is that Samsung did produce 14 nm chips ahead of intel . Intel may have been selling them ahead of samsung but that seems to be more a claim for marketing purposes . Did anyone actually manage to buy a broadwell cpu back when they were "released" ?

As for "low" power and "high" power variants . Its not a big deal . The "high" LPP variant has been in mass production for more than 6 months now at GloFo . Its an established process and wont be problematic for Zen production or for the new graphics chips likely to be produced .
It seems there are no technical reasons to suspect Zen will be delayed . There is only the time it takes to establish a production line for the new product .

And mousemonkey I dont see AMD fanatics anywhere here . For instance of the 3 pc's and 1 laptop I own 3 of them have intel processors . The forth has an AMD . It was the best most cost effective choice.

There do seem to be some people whose view of the world begins and ends at intel though . Even to the point they feel the need to try and insult anyone who suggests AMD are a viable alternative

 

[photonboy]

i would like to quell any feelings of AMD's complete failure. Intel nor nvidia will let it happen. the risk and consequences of being a monopoly is far greater than having a competitor that is barely surviving.

How exactly would NVidia prevent AMD from failing?

Sure, there are laws about monopolies and of course competition is a good thing but it's not like NVidia's going to give them a Billion USD to keep afloat.

We have regulation called "competition law" to regulate anti-competitive conduct by companies but it doesn't prevent NVidia from becoming a monopoly if AMD fails, it simply attempts to prevent them failing because NVidia did something illegal.

*And why exactly would NVidia not want to be a monopoly? That's always a good thing for a company (not consumers).

 

[IInuyasha74]

i would like to quell any feelings of AMD's complete failure. Intel nor nvidia will let it happen. the risk and consequences of being a monopoly is far greater than having a competitor that is barely surviving.

How exactly would NVidia prevent AMD from failing?

Sure, there are laws about monopolies and of course competition is a good thing but it's not like NVidia's going to give them a Billion USD to keep afloat.

We have regulation called "competition law" to regulate anti-competitive conduct by companies but it doesn't prevent NVidia from becoming a monopoly if AMD fails, it simply attempts to prevent them failing because NVidia did something illegal.

*And why exactly would NVidia not want to be a monopoly? That's always a good thing for a company (not consumers).

This is true, but just to continue the discussion, it should be pointed out Nvidia won't have to worry about being a monopoly. If AMD went out of business, there is a good chance someone would buy off the graphics tech to continue in some way. Even if they didn't, they still will compete against Intel. They won't have any competition in the high-end but they will in the low-end and possibly one day in the low-mid range from Intel. They will also be able to point to all the mobile GPU developers in order to say they aren't a monopoly, so they won't have any issue there.

Intel will be similar. They won't have any real competition for consumer PCs, but with IBM, ARM, Qualcomm, Nvidia and Apple still technically producing their own CPUs and all the SoC makers producing their hardware and going into devices, it will probably be enough to dodge being called a monopoly.

 

[jimmysmitty]

Another thing mentioned in the article that is misreported is possible technical difficulties with the process node .
Its actually more mature than intels 14 nm . Samsung are licensing the tech to Global Foundries , and Samsung have been at 14 nm longer than Intel have .

Its very hard to find something that is factual in the article but I will presume the authors name is correct

There is just so much wrong here.

Samsungs first 14nm chip was tapped out in 2012 and it was just a test chip. Their first actual chip was not released until 2015 with the Galaxy S6 line.

Intel had 14nm tapped out in 2011 and started building their FAB the same year, FAB 42 in Chandler. Intels first chips with 14nm was demoed in 2013, a Ultrabook with an actual working product not a test chip, and was in full shipping production in 2014.

To top it off, Samsungs 14nm is not geared towards high power chips, that has to be tweaked by GloFlo for that purpose while Intels 14nm is geared towards high power chips. Yields will easily be better for Samsung with low power, more simplistic CPUs vs a higher power and more complex CPU. AMD will not get the same yields out of Samsungs 14nm as Samsung does.

Dude, you're dealing with AMD fanatics so why would you expect anything less? :lol:

The point of course is that Samsung did produce 14 nm chips ahead of intel . Intel may have been selling them ahead of samsung but that seems to be more a claim for marketing purposes . Did anyone actually manage to buy a broadwell cpu back when they were "released" ?

As for "low" power and "high" power variants . Its not a big deal . The "high" LPP variant has been in mass production for more than 6 months now at GloFo . Its an established process and wont be problematic for Zen production or for the new graphics chips likely to be produced .
It seems there are no technical reasons to suspect Zen will be delayed . There is only the time it takes to establish a production line for the new product .

And mousemonkey I dont see AMD fanatics anywhere here . For instance of the 3 pc's and 1 laptop I own 3 of them have intel processors . The forth has an AMD . It was the best most cost effective choice.

There do seem to be some people whose view of the world begins and ends at intel though . Even to the point they feel the need to try and insult anyone who suggests AMD are a viable alternative

What 14nm chip did Samsung have before Intel again? The first was the Exynos in the Galaxy S6, a 2015 product, while Intel demoed a 14nm CPU in 2013 and launched them in the Core M series in 2014. And again, Samsung tapped their 14nm in 2013 while Intel tapped theirs out in 2011 (hence why they started building FAB 42). Everything I see shows Intel had 14nm first.

And high power or low power does matter. Samsungs 14nm LP is very mature, much like Intels 14nm is now, but the tweaks to it GF will have made will affect yield rates as will the CPU design itself.

Be very careful though. The fact that you seem to think there are Intel fanatics but not AMD shows a lot. There are both. You might not see them but I sure as hell do.

 

[Outlander_04]

The LPP version is the high power version of Samsungs 14 nm finfets. As I said its been in production for over 6 months now at GloFo .
Technically it will not be a challenge . Yield rates on wafers are already good enough .
And with AMD staying with the 4/6/8 core approach its likely that they will be maximizing returns from wafers from very early in the manufacturing cycle .

Yes there are people who prefer AMD . So what? Some people like Ford , some like Chevy . Is one choice wrong?
Why do some ford buyers keep insisting that Chevy is not an option despite plenty of evidence to the contrary?
Why did a moderator on these forums insist I was out of line for publishing benchmarks that showed Chevys worked?

As for the "delay" suggested in this article . Is there any basis for that claim? Any new report from manufacturer?
If there is no reason to suspect a delay at all why report that there is?

 

[jimmysmitty]

The 14nm LPP is still a low power 14nm, it is a performance version. It is not a high power version, it uses the same Core VDD of 0.8v as the 14nm LPE. It is by default geared for low power designs, not for high powered designs. It has to be tweaked by GF for high power designs and thus it will no longer be as mature as it will be for the low power designs that have been running on it for the bast 6 months.

And please stay on topic. That has nothing to do with what we were discussing. I was saying that you were wrong, Samsung has not had 14nm longer than Intel.

http://seekingalpha.com/article/3741456-advanced-micro-devices-amd-at-raymond-james-technology-investors-conference-transcript?page=4

That is the linked article which states that a top person at AMD said 12-14 months is the usual time for shipping to OEM channels from a process being tapped out.

As for GF having it 6 months ago, I am not 100% sure it was. Samsung just launched their 14nm CPU in the Galaxy S6 6 months ago. I would say that AMD tapped the designs more recently, maybe within the last 2-3 months. If so then 12-14 months puts launch of Zen in late 2016 to early 2017 so long as yields are good.

 

[Outlander_04]

And staying on topic Q4 2016 is the time that Zen is expected in retail channels . Which means there is no delay

 

[jimmysmitty]

And staying on topic Q4 2016 is the time that Zen is expected in retail channels . Which means there is no delay

What is your source for this? The article has a linked source to its information.

 

[Outlander_04]

And staying on topic Q4 2016 is the time that Zen is expected in retail channels . Which means there is no delay

What is your source for this? The article has a linked source to its information.

Did you read that link yourself?
It does not mention a delay
It does say volume production of 14 nm finfet products in 2016
It also says that high performance desktop Zen is first , and that the revenue ramp up will start in 2016

 

[jimmysmitty]

And staying on topic Q4 2016 is the time that Zen is expected in retail channels . Which means there is no delay

What is your source for this? The article has a linked source to its information.

Did you read that link yourself?
It does not mention a delay
It does say volume production of 14 nm finfet products in 2016
It also says that high performance desktop Zen is first , and that the revenue ramp up will start in 2016

We have taped out multiple products just in the last few months on FinFET and we will have volume production in FinFET products in 2016 at more foundry partners

That specifically says they tapped the products out in the last few months and that volume production will be in 2016. That does not mean shipping.

We haven’t given the specifics but you know they taped out in the last few months and typically what happens on the product depending on when you taped it out it can be 12 to 14 months from the time you tape out the products to when you start shipping the products

He specifically states again that products (most likely Zen) were tapped out in the last few months and that from that point it is normally 12-14 months before they start to ship the products to OEM channels. If we do what few normally means (2-3) then the earliest we might see Zen start to ship is 10/2016. The latest would be 12/2016-1/2017.

This does not say anywhere that it is set for a Q4 2016 release in fact from the same paragraph:

but we haven’t been specific about you know exact timeframe

There is no specific time frame for the product to be shipped right now, at least they have not given us one yet. And there are a lot of factors that will matter such as yields, I know you think that somehow GloFlo will have a more mature 14nm process node than Intel and have great yields but they wont. There is a lot of reasons why. Again they are taking a process and changing it to work better with their specific design (the 14nm LPP is not a high power process, it is a low power process). That will lower yields. Then we have Zen, a new uArch, which alone will affect yields.

Nowhere in the TH article did the author state there is going to be a delay but that there might possibly be a delay. Possibly means it may or may not but with the information in the article it looks like Zen might not roll out until Q1 2017, again not a 100% certainty but it is possible. They could delay it for financial reasons, sometimes it is best to drop it later to have a better earnings report for a specific quarter.

Again I will ask you, what sources do you have that state 100% absolutely that Zen is slated for full volume production and release in Q4 2014 that can refute what this AMD executive has stated?

 

[Outlander_04]

Actually the revenue ramp cant start until after products are shipped . Revenue ramp in 2016 can only mean one thing

But since you have just said that there is no specific time frame given HOW COULD THERE POSSIBLY BE CONSIDERED TO BE DELAYED ?
LOL
Would you agree that an equally valid title backed by just as many facts could have been :
"New AMD processors smash intel performance"
There is just as much evidence for that as anything else at this point

And once again there is no technical or yield problem with the 14nm lpp process . None what so ever .

 

[jimmysmitty]

Actually the revenue ramp cant start until after products are shipped . Revenue ramp in 2016 can only mean one thing

But since you have just said that there is no specific time frame given HOW COULD THERE POSSIBLY BE CONSIDERED TO BE DELAYED ?
LOL
Would you agree that an equally valid title backed by just as many facts could have been :
"New AMD processors smash intel performance"
There is just as much evidence for that as anything else at this point

And once again there is no technical or yield problem with the 14nm lpp process . None what so ever .

A delay to 2017 would be a delay but again the article stated it was possible not a 100%.

And again you keep saying things with no sources to back anything you are saying up. How would you know the yield rates of GloFlos 14nm LPP? Do you work for them?

And do you think if there was they would say anything publicly about it? That would hurt business and sales. Look what happened to TSMC and 20nm. Both AMD and nVidia skipped it due to yield and technical problems.

 

[Outlander_04]

Actually the revenue ramp cant start until after products are shipped . Revenue ramp in 2016 can only mean one thing

But since you have just said that there is no specific time frame given HOW COULD THERE POSSIBLY BE CONSIDERED TO BE DELAYED ?
LOL
Would you agree that an equally valid title backed by just as many facts could have been :
"New AMD processors smash intel performance"
There is just as much evidence for that as anything else at this point

And once again there is no technical or yield problem with the 14nm lpp process . None what so ever .

A delay to 2017 would be a delay but again the article stated it was possible not a 100%.

And again you keep saying things with no sources to back anything you are saying up. How would you know the yield rates of GloFlos 14nm LPP? Do you work for them?

And do you think if there was they would say anything publicly about it? That would hurt business and sales. Look what happened to TSMC and 20nm. Both AMD and nVidia skipped it due to yield and technical problems.

The LPE and LPP processes are structurally identical . Probably what has changed is one of the metal gates .This may be due to nothing more than licensing arrangements to a patent holder. Its not harder to make one than the other . The LPE version may exist purely as a cost saving because it works fine for its intended use without paying that royalty .
There is no issue with production because they are already manufacturing these devices .

On top of that AMD have always maximized sales yields by binning and selling as 2,3,4 cores with athlon and then 4,6,8 with FX [ because of the paired cores ] . The relative prices of 6 and 8 core variants is all that will be affected by yields if history repeats .

Historically there is a website that has pretty much nailed every move AMD has made .
[ and its not tomshardware that even editorialized that AMD had given up top end processors . LOL ]
Most likely the best public information regarding AMD in 2016 can be found here
http://wccftech.com/amd-zen/
http://wccftech.com/amd-zen-launch-q4-2016/
http://wccftech.com/amd-zen-k12-taped/
The information is very consistent with the link in this article at toms .

 

[jimmysmitty]

Actually the revenue ramp cant start until after products are shipped . Revenue ramp in 2016 can only mean one thing

But since you have just said that there is no specific time frame given HOW COULD THERE POSSIBLY BE CONSIDERED TO BE DELAYED ?
LOL
Would you agree that an equally valid title backed by just as many facts could have been :
"New AMD processors smash intel performance"
There is just as much evidence for that as anything else at this point

And once again there is no technical or yield problem with the 14nm lpp process . None what so ever .

A delay to 2017 would be a delay but again the article stated it was possible not a 100%.

And again you keep saying things with no sources to back anything you are saying up. How would you know the yield rates of GloFlos 14nm LPP? Do you work for them?

And do you think if there was they would say anything publicly about it? That would hurt business and sales. Look what happened to TSMC and 20nm. Both AMD and nVidia skipped it due to yield and technical problems.

The LPE and LPP processes are structurally identical . Probably what has changed is one of the metal gates .This may be due to nothing more than licensing arrangements to a patent holder. Its not harder to make one than the other . The LPE version may exist purely as a cost saving because it works fine for its intended use without paying that royalty .
There is no issue with production because they are already manufacturing these devices .

On top of that AMD have always maximized sales yields by binning and selling as 2,3,4 cores with athlon and then 4,6,8 with FX [ because of the paired cores ] . The relative prices of 6 and 8 core variants is all that will be affected by yields if history repeats .

Historically there is a website that has pretty much nailed every move AMD has made .
[ and its not tomshardware that even editorialized that AMD had given up top end processors . LOL ]
Most likely the best public information regarding AMD in 2016 can be found here
http://wccftech.com/amd-zen/
http://wccftech.com/amd-zen-launch-q4-2016/
http://wccftech.com/amd-zen-k12-taped/
The information is very consistent with the link in this article at toms .

No the LPE version exists because Samsung wanted to get their 14nm out earlier and the LPP was their original design. The LP literally stands for "Low Power" while the E means "Early" and the P means "Performance". I am not arguing that the LPE and LPP are similar. I am stating that GloFlow is not using the base LPP for Zen CPUs but they are having to modify it to be able to work better with a high power design. And it is not as simple as changing a gate, there is a lot to a process node. Believe me or not, the community college I went to had a scholarship through Intel and worked with Intel very closely. A friend of mine did the scholarship and now works for Intel in Chandler (we were in a city 2 hours away). I have personally seen what it takes to create and fabricate a chip and it is not some simple task like turning a gate on/off.

If it were than easy then Intel would have moved to 16nm sooner than 14nm. However that specific node is design for NAND and SRAM, not for CPUs and high power processing cores. You are trying to simplify what takes years of work to do.

Binning is not new. You used to be able to do a pencil mod to the Radeon 9500 to make it a 9700 Pro or to the Q6600 to make its FSB default to 333MHz instead of having to tweak the BIOS. Every company does it. Profits are not the only thing affected by yield, if yields are too low and they can't fill OEM channels they will suffer more than if they delay and can fill channels.

Are you honestly linking WCCFTech? They are nothing but a rumor mill. They spit out rumors left and right even if the rumors are known to be bogus.

http://wccftech.com/amd-r9-390x-8-gb-hbm/ - Wrong, 4GB limitation to HBM
http://wccftech.com/fiji-xt-4gb-amd-dual-8gb-hbm-msrp-849/ - Wrong, again 4GB limitation and MSRP was $650

It is really hard to trust a website that reports daily on rumors just to get hits.

So you have an actual source or is the rumor mill your only source?

 

[Mousemonkey]

Actually the revenue ramp cant start until after products are shipped . Revenue ramp in 2016 can only mean one thing

But since you have just said that there is no specific time frame given HOW COULD THERE POSSIBLY BE CONSIDERED TO BE DELAYED ?
LOL
Would you agree that an equally valid title backed by just as many facts could have been :
"New AMD processors smash intel performance"
There is just as much evidence for that as anything else at this point

And once again there is no technical or yield problem with the 14nm lpp process . None what so ever .

A delay to 2017 would be a delay but again the article stated it was possible not a 100%.

And again you keep saying things with no sources to back anything you are saying up. How would you know the yield rates of GloFlos 14nm LPP? Do you work for them?

And do you think if there was they would say anything publicly about it? That would hurt business and sales. Look what happened to TSMC and 20nm. Both AMD and nVidia skipped it due to yield and technical problems.

The LPE and LPP processes are structurally identical . Probably what has changed is one of the metal gates .This may be due to nothing more than licensing arrangements to a patent holder. Its not harder to make one than the other . The LPE version may exist purely as a cost saving because it works fine for its intended use without paying that royalty .
There is no issue with production because they are already manufacturing these devices .

On top of that AMD have always maximized sales yields by binning and selling as 2,3,4 cores with athlon and then 4,6,8 with FX [ because of the paired cores ] . The relative prices of 6 and 8 core variants is all that will be affected by yields if history repeats .

Historically there is a website that has pretty much nailed every move AMD has made .
[ and its not tomshardware that even editorialized that AMD had given up top end processors . LOL ]
Most likely the best public information regarding AMD in 2016 can be found here
http://wccftech.com/amd-zen/
http://wccftech.com/amd-zen-launch-q4-2016/
http://wccftech.com/amd-zen-k12-taped/
The information is very consistent with the link in this article at toms .

No the LPE version exists because Samsung wanted to get their 14nm out earlier and the LPP was their original design. The LP literally stands for "Low Power" while the E means "Early" and the P means "Performance". I am not arguing that the LPE and LPP are similar. I am stating that GloFlow is not using the base LPP for Zen CPUs but they are having to modify it to be able to work better with a high power design. And it is not as simple as changing a gate, there is a lot to a process node. Believe me or not, the community college I went to had a scholarship through Intel and worked with Intel very closely. A friend of mine did the scholarship and now works for Intel in Chandler (we were in a city 2 hours away). I have personally seen what it takes to create and fabricate a chip and it is not some simple task like turning a gate on/off.

If it were than easy then Intel would have moved to 16nm sooner than 14nm. However that specific node is design for NAND and SRAM, not for CPUs and high power processing cores. You are trying to simplify what takes years of work to do.

Binning is not new. You used to be able to do a pencil mod to the Radeon 9500 to make it a 9700 Pro or to the Q6600 to make its FSB default to 333MHz instead of having to tweak the BIOS. Every company does it. Profits are not the only thing affected by yield, if yields are too low and they can't fill OEM channels they will suffer more than if they delay and can fill channels.

Are you honestly linking WCCFTech? They are nothing but a rumor mill. They spit out rumors left and right even if the rumors are known to be bogus.

http://wccftech.com/amd-r9-390x-8-gb-hbm/ - Wrong, 4GB limitation to HBM
http://wccftech.com/fiji-xt-4gb-amd-dual-8gb-hbm-msrp-849/ - Wrong, again 4GB limitation and MSRP was $650

It is really hard to trust a website that reports daily on rumors just to get hits.

So you have an actual source or is the rumor mill your only source?

Zen is still too far out in time for anyone to accurately pinpoint an exact release date, at this point in time it would be very challenging even for AMD to commit to a specific date internally. Still, we’ve learned that AMD’s planning to debut Zen based desktop FX CPUs by Q4 2016, just in time for next year’s holiday season. While Zen based APUs are expected to debut at a later date in 2017.

It’s important to emphasize that AMD has not yet announced any specific release dates and the company’s official statement is still “on track for ‘Zen‘ availability in 2016 with first full year of revenue in 2017″. So while we’ve managed to confirm what AMD’s plans are for today it would be irresponsible not to draw attention to the fact that these plans could change in the future. But so far it’s been made very clear to us that late 2016 and beyond is AMD’s current target.

It seems to me as though AMD are hedging their bets that they can deliver in 2016 from they way that article is written. :lol:

 

[Mousemonkey]

I wonder how long before Outlander_04 and alextheblue downvote that post? :lol:

 

[Outlander_04]

I wonder how long before Outlander_04 and alextheblue downvote that post? :lol:

Your post had so much entertainment value I cant bring my self to vote it down . Im still laughing

 

[Outlander_04]

Actually the revenue ramp cant start until after products are shipped . Revenue ramp in 2016 can only mean one thing

But since you have just said that there is no specific time frame given HOW COULD THERE POSSIBLY BE CONSIDERED TO BE DELAYED ?
LOL
Would you agree that an equally valid title backed by just as many facts could have been :
"New AMD processors smash intel performance"
There is just as much evidence for that as anything else at this point

And once again there is no technical or yield problem with the 14nm lpp process . None what so ever .

A delay to 2017 would be a delay but again the article stated it was possible not a 100%.

And again you keep saying things with no sources to back anything you are saying up. How would you know the yield rates of GloFlos 14nm LPP? Do you work for them?

And do you think if there was they would say anything publicly about it? That would hurt business and sales. Look what happened to TSMC and 20nm. Both AMD and nVidia skipped it due to yield and technical problems.

The LPE and LPP processes are structurally identical . Probably what has changed is one of the metal gates .This may be due to nothing more than licensing arrangements to a patent holder. Its not harder to make one than the other . The LPE version may exist purely as a cost saving because it works fine for its intended use without paying that royalty .
There is no issue with production because they are already manufacturing these devices .

On top of that AMD have always maximized sales yields by binning and selling as 2,3,4 cores with athlon and then 4,6,8 with FX [ because of the paired cores ] . The relative prices of 6 and 8 core variants is all that will be affected by yields if history repeats .

Historically there is a website that has pretty much nailed every move AMD has made .
[ and its not tomshardware that even editorialized that AMD had given up top end processors . LOL ]
Most likely the best public information regarding AMD in 2016 can be found here
http://wccftech.com/amd-zen/
http://wccftech.com/amd-zen-launch-q4-2016/
http://wccftech.com/amd-zen-k12-taped/
The information is very consistent with the link in this article at toms .

No the LPE version exists because Samsung wanted to get their 14nm out earlier and the LPP was their original design. The LP literally stands for "Low Power" while the E means "Early" and the P means "Performance". I am not arguing that the LPE and LPP are similar. I am stating that GloFlow is not using the base LPP for Zen CPUs but they are having to modify it to be able to work better with a high power design. And it is not as simple as changing a gate, there is a lot to a process node. Believe me or not, the community college I went to had a scholarship through Intel and worked with Intel very closely. A friend of mine did the scholarship and now works for Intel in Chandler (we were in a city 2 hours away). I have personally seen what it takes to create and fabricate a chip and it is not some simple task like turning a gate on/off.

If it were than easy then Intel would have moved to 16nm sooner than 14nm. However that specific node is design for NAND and SRAM, not for CPUs and high power processing cores. You are trying to simplify what takes years of work to do.

Binning is not new. You used to be able to do a pencil mod to the Radeon 9500 to make it a 9700 Pro or to the Q6600 to make its FSB default to 333MHz instead of having to tweak the BIOS. Every company does it. Profits are not the only thing affected by yield, if yields are too low and they can't fill OEM channels they will suffer more than if they delay and can fill channels.

Are you honestly linking WCCFTech? They are nothing but a rumor mill. They spit out rumors left and right even if the rumors are known to be bogus.

http://wccftech.com/amd-r9-390x-8-gb-hbm/ - Wrong, 4GB limitation to HBM
http://wccftech.com/fiji-xt-4gb-amd-dual-8gb-hbm-msrp-849/ - Wrong, again 4GB limitation and MSRP was $650

It is really hard to trust a website that reports daily on rumors just to get hits.

So you have an actual source or is the rumor mill your only source?

My ex-wifes brother is a chip designer working mostly in the defense industry . One of my clients is a chip designer working on controller systems . I have a little insight from these people

I think we could debate this all day . You are speculating that AMD can not meet their supposed timeline . I feel its much more likely that they can . Especially because GloFo are already using the LPP process

Why dont we meet back here in Q4 2016 . One of us is bound to be wrong 😉