I wonder if this will eventually be used for console chips, allowing MS, Nintendo and Sony to create their slimline cheaper console versions?

inb4 PC Gamer Master Race rage against me. My girlfriend wants a Wii U, not paying £230 for a console so she can play only Mario Kart.

RCoonI wonder if this will eventually be used for console chips, allowing MS, Nintendo and Sony to create their slimline cheaper console versions?

inb4 PC Gamer Master Race rage against me. My girlfriend wants a Wii U, not paying £230 for a console so she can play only Mario Kart.

Yeah, AMD kept optical-shrinking its Xbox360 chip down from 65 nm down to 55 nm, and 40 nm. It's conceivable that its 28 nm SoCs could be shrunk down to 14 nm.

Yes but Jim Keller has since left the company....again

btarunrYeah, AMD kept optical-shrinking its Xbox360 chip down from 65 nm down to 55 nm, and 40 nm. It's conceivable that its 28 nm SoCs could be shrunk down to 14 nm.

IBM, right? AMD only provided the Xbox360 GPU. The Xenos GPU went from 90, to 65, to 45nm.

This is very exciting news because AMD will be:
1) nipping at the heels of Intel again.
2) have a 2 nm advantage on NVIDIA's TSMC production.
3) AMD can compete in mobile/low power products again.

robert3892Yes but Jim Keller has since left the company....again

Yeah, after finishing off the operative portion of Zen's development. It's not as if Zen is in jeopardy with his departure.

Basically Jim Keller is a design mercenary. He gets paid, leads the team to design a product and then leaves. I'm sure AMD would employ him full time if they were financially better situated, but this is how it is.

I just always find it funny how companies always find money for suit wearing briefcase carrying men with ties and shiny shoes but never have the money for engineers who make products which fuel the whole operation. Sure marketing is also a big part of it, but you can't sell hot water as the best thing after sliced bread. You need solid products that you can then use to brag about. If boards would value workers below them more, I'm certain companies would operate better product and financially wise.

I have to admit... those are rather bad ass selfies...

Ferrum MasterI have to admit... those are rather bad ass selfies...

Only one that tops it is this one:



Which means it's pretty damn high on the cool scale :)

This news came out only days, even just hours, after rumors that AMD was going to TSMC for about everything, Zen and GPUs.

john_This news came out only days, even just hours, after rumors that AMD was going to TSMC for about everything, Zen and GPUs.

Yeah I dread the thought of AMD signing with TSMC, just before Global Foundries showed good yields. Crossing my fingers here.

john_This news came out only days, even just hours, after rumors that AMD was going to TSMC for about everything, Zen and GPUs.

Bear in mind that:
1. This news originates from Globalfoundries. Their record on public pronouncements of their yields in the past haven't exactly tallied with what actually happened (see Llano, Bulldozer et al)., and
2. Zen, I believe, was generally supposed to be a GloFo 14nm product. 14nm LPE wasn't an option by the sounds of it, andindustry speculation was that 14nm LPP yields weren't great. I don't think 14nm LPE/LPP was ever a serious contender for GPUs (at least not the large die ones) since the process (as GloFo themselves are stating) is geared towards lower power applications.

If it isn't suitable for CPUs and GPUs, that bodes bad for AMD.

HumanSmokeBear in mind that:
1. This news originates from Globalfoundries. Their record on public pronouncements of their yields in the past haven't exactly tallied with what actually happened (see Llano, Bulldozer et al)., and
2. Zen, I believe, was generally supposed to be a GloFo 14nm product. 14nm LPE wasn't an option by the sounds of it, andindustry speculation was that 14nm LPP yields weren't great. I don't think 14nm LPE/LPP was ever a serious contender for GPUs (at least not the large die ones) since the process (as GloFo themselves are stating) is geared towards lower power applications.

What do you make up from my post? That I take for granted what GF came out and said in their announcement? IF AMD is in fact going at TSMC, GF's press release is just damage control because many will think the worst looking at the most loyal customer abandoning GF. Expectations where high when they announced their cooperation with Samsung. And the higher you go, the bigger the fall.

Wasn't there a report recently that AMD switched back to TSMC because they were fed up with Gloflo? That Zen had been delayed due to it but were back on schedule with TSMC? Thought I saw that somewhere recently. Maybe it was elsewhere. Could have just been a rumor.

I'd be glad if they did because things were looking like 2017 for Zen instead of 16.

john_What do you make up from my post? That I take for granted what GF came out and said in their announcement?

Not in the least. This announcement concerns GloFo's 14nm plans and Zen - and probably the smaller GPUs of the Arctic Islands series. You made the statement that you thought TSMC were rumoured to be taking all AMD's business - which I have not seen evidence of, and don't believe to be true.
All I stated was that both GloFo AND TSMC will both probably have a share of the business, and GloFo's press releases may not reflect actuality in regards process maturity- which might account for any rumours you may have seen regarding TSMC having all AMD's business (although I have seen no credible information to say this is so). To my mind, these can (and probably are) mutually exclusive, which is why deliberately itemized them.
GloFo's timetable and yield need have no impact on whether it gets AMD's business - the only variable is the timeframe for production of the parts.

Any other inference is yours.

NC37Wasn't there a report recently that AMD switched back to TSMC because they were fed up with Gloflo? That Zen had been delayed due to it but were back on schedule with TSMC?

Can't say I've seen that - at least not from a creditable source. AFAIK, the rumour/report was that Samsung and GloFo had production ramp issues that had two knock-on effects - Zen's introduction has been put back to early 2017 (announced by AMD themselves), and Apple diverting a greater share of A9 production to TSMCfrom GloFo and Samsung because of their concerns, which led to TSMC securing the A10 contract.

HumanSmokeYou made the statement that you thought TSMC were rumoured to be taking all AMD's business - which I have not seen evidence of, and don't believe to be true.

Try not to play with words and try not to confuse "thoughts" with "evidence" and "rumors" in a way that makes it look like we are in a court room and the choice of words is the most important thing in the world.

amd TSMC - Αναζήτηση Google

HumanSmokeBear in mind that:
1. This news originates from Globalfoundries. Their record on public pronouncements of their yields in the past haven't exactly tallied with what actually happened (see Llano, Bulldozer et al)., and
2. Zen, I believe, was generally supposed to be a GloFo 14nm product. 14nm LPE wasn't an option by the sounds of it, andindustry speculation was that 14nm LPP yields weren't great. I don't think 14nm LPE/LPP was ever a serious contender for GPUs (at least not the large die ones) since the process (as GloFo themselves are stating) is geared towards lower power applications.

Could you elaborate on LPE vs LPT vs what Intel vs maybe TSMC vs maybe Samsung has pretty please?
I hope I don't ask for too much.
=)

Anal retentive questions here.

1) What exactly is "good yields?" The human part of me wants to say reasonably high, but the PR side of me says that it's anything in the ballpark of what they were expecting. If that was 80%, and they actually only got 75%, it's not a "good yield" by the 25% loss but it's about the 80% they were expecting.

2) Closer to accepting orders isn't closer to getting an in control and volume producing run. These test runs are limited batch quantities, intelligent workers, and highly controlled environments. None of this reflects day to day operations, where sometimes people do stupid things.

3) Global Foundries is always late. Despite having a whole generation where there was no shrink they're just now getting this stuff in order. You'll excuse me, but volume production by late 2016 sounds like they're pushing it. Samples still need to be analyzed, manufacturing still needs to have time to adopt the process, and AMD is going to have to do some subsequent design work. This definitely sounds like PR/damage control, rather than a triumph.


This said, it'd be nice if Global Foundries got their stuff together. Hopefully this will give AMD an advantage in the GPU market, and bring them back to the table with CPUs. I'm not holding my breath for the latter, but the former may be how AMD can actually start digging themselves out of their current hole.

john_Try not to play with words and try not to confuse "thoughts" with "evidence" and "rumors" in a way that makes it look like we are in a court room and the choice of words is the most important thing in the world.

amd TSMC - Αναζήτηση Google

The first two links...WTFtech ( Where imbeciles confront morons in an endless battle to prove which group has the lowest IQ), and Fudzilla ( nuff said)
Third link PCGH - which actually analysed what was said by expreview ( which in turn came from some obscure Italian site - Bits and Chips) rather than what was conjured out of thin air:

Contrary to these rumors, however, GlobalFoundries recently announced that the 14-nanometer process went even beyond the previous expectations and plans and you can work confidently with the new chips.Soon, the process should go into mass production.Whether AMD should therefore have actually changed the chip maker, has so far not officially confirmed - and sounds not very credible, given the consequent time problems with the conversion.TSMC is expected to continue but finished graphics chips for AMD.

As I said - I haven't seen any evidence that AMD has moved Zen production from GloFo to TSMC, and don't believe it to be the case. You on the other hand may see it differently, but if you're trying to convince me, you'll need to do better than a 2-second Google search which brings up the usual clickbait suspects all getting their info from a single unsubstantiated rumour.

Sorry if you have a distaste for the English language used correctly. If it causes you so much angst maybe you can just skip more of the words in my posts (which you tend to do in any case), or just stop reading them altogether. Problem solved.

medi01Could you elaborate on LPE vs LPT vs what Intel vs maybe TSMC vs maybe Samsung has pretty please?
I hope I don't ask for too much.
=)

14nm LPE and LPP are the same process. The E in LPE stands for "Early" and as such isn't as well optimized for power, voltage leakage etc. LPP is the mature (and later) version. As for FinFET process nodes, Google is your friend for general articles. This one looks like a decent introductory primer. For a more in depth analysis of foundry processes I would suggest SemiWiki that provides good non-partisan overviews of the industry.

HumanSmokeThe first two links...WTFtech ( Where imbeciles confront morons in an endless battle to prove which group has the lowest IQ), and Fudzilla ( nuff said)
Third link PCGH - which actually analysed what was said by expreview ( which in turn came from some obscure Italian site - Bits and Chips) rather than what was conjured out of thin air:

As I said - I haven't seen any evidence that AMD has moved Zen production from GloFo to TSMC, and don't believe it to be the case. You on the other hand may see it differently, but if you're trying to convince me, you'll need to do better than a 2-second Google search which brings up the usual clickbait suspects all getting their info from a single unsubstantiated rumour.

Sorry if you have a distaste for the English language used correctly. If it causes you so much angst maybe you can just skip more of the words in my posts (which you tend to do in any case), or just stop reading them altogether. Problem solved.

14nm LPE and LPP are the same process. The E in LPE stands for "Early" and as such isn't as well optimized for power, voltage leakage etc. LPP is the mature (and later) version. As for FinFET process nodes, Google is your friend for general articles. This one looks like a decent introductory primer.

Do you know what a forum is? Let me explain to you a couple things with my poor English. In a forum you don't only post evidence, you don't post only to convince someone, it's not a court room and definitely not an arena. So people will also post things they just hear/read about, say their opinions and just try to have a polite conversation like in a cafe or something. So, would you PLEASE stop trying to interpret me in a convenient way, rushing to show me where I am wrong in about everything? Find a hobby. Another hobby, not my posts.

john_So, would you PLEASE stop trying to interpret me in a convenient way, rushing to show me where I am wrong in about everything?

Who said you were wrong? All I said was that there are conflicting reports - as you yourself pointed out- and of those reports, gave an opinion on what I thought was the most creditable.

You seem to be nurturing a persecution complex....all because of what? Because we both made public predictions that I said I would revisit? What is the point of making predictions if you don't want them found to be true or not?
Do yourself a favour - don't read my postings. I'd hate to be the cause of you keeling over from hypertension.

Eh?, No-one is the slightest creeped out by the two bunny suits staring deep into each others eyes. That 14nm process must be something else.

HumanSmoke
Don't do like you don't understand. First post "bear in mind", so I missed something, then "You made the statement that you thought", "rumours you may have seen" "Any other inference is yours.". Should I add here the results of your psychological examination about me, my distaste for the English language, the angst, and me nurturing a persecution complex. And you shouldn't have problem if I am not.... reading all the words when you conveniently bypass every part of my posts you prefer to ignore. I have no problem if you "gave an opinion on what you thought was the most creditable", but try leaving me out of your opinion.

Anyway, they are just rumors and a PR from a company about itself. Nothing can taken as truth or even close to truth. Thing like that we keep them in the back of our heads and just wait to see if confirmations will come out in the future. For now we can only speculate and that's all. It's too early to conclude, or even guess what speculations, could be closer to reality.

PS Going to make a lemonade now for my hypertension that you diagnosed over the internet.

john_This news came out only days, even just hours, after rumors that AMD was going to TSMC for about everything, Zen and GPUs.

Exactly, I see this as "speak" to quell bad press, though faithfully true. The news has for some reason been about TSMC having engineering samples shipped, or then a rumor that Zen was going to TSMC, because GloFo funding was pulled due to low oil revenues. It's almost like someone plants these to prompt GloFo into a response. And I’m not saying it’s companies (like Intel or Nvidia), more just other there’s a lot interested parties and investors wanting an "inkling" of what the heck is up at GloFo? Whatever the reason GloFo had rather holding information as to 14FinFET close to their chest, though finally the chatter necessitates a counter to quell rumors. Where those rumor correct… we wait and given what’s on-the-line for all involved it will be a turbulent year.

CasecutterThe news has for some reason been about TSMC having engineering samples shipped, or then a rumor that Zen was going to TSMC, because GloFo funding was pulled due to low oil revenues. It's almost like someone plants these to prompt GloFo into a response.

Well, the original rumour about AMD switching to TSMC from GloFo came from andItalian tech site whose contributor is so confident that the rumour is reality that he is quite content to allow Expreview (who reprinted the rumour and most attribute as the source) to remain the focus of the information.

CasecutterAnd I’m not saying it’s companies (like Intel or Nvidia), more just other there’s a lot interested parties and investors wanting an "inkling" of what the heck is up at GloFo?

Oddly enough, the site that initiated the rumour is avidly AMD orientated rather that Intel or Nvidia, and pretty much stated that AMD related news gets the page hits for them:

At Bits And Chips the news about AMD are the most read. Idem at TechPowerUp!, WCCFTech and other sites. AMD is not the most important company, but it moves a lot of visits.

Small time tech site struggling for page views and ad revenue in a market very competitive for readership initiates a rumour guaranteed to raise readership. Seems like smoking gun material, no?

CasecutterWhatever the reason GloFo had rather holding information as to 14FinFET close to their chest, though finally the chatter necessitates a counter to quell rumors.

GloFo have been through this before. Remember when the company were going through their "our yields are great" phase for 32nm? which then morphed into "yeah, maybe there was a problem - but its fixed now" and then "Yeah, it's really screwed up"
Add in the cancelled processes, the late ramp and bad yields of those it has undertaken, and you can see why GloFo probably aren't too forthcoming with information.

XzibitEh?, No-one is the slightest creeped out by the two bunny suits staring deep into each others eyes. That 14nm process must be something else.

They probably share adjoining work stations in that cozy environment you can see in the background.