TSMC to start 16nm volume production in 1Q15, says report

[Ayahuasca]

Seems to only be for their Tegra chip though.

 

[Rexton]

Are AMD still aiming for 20nm by 2015 and 16nm by 2016? Or have they decided on 14nm?

I'm sure I read somewhere they were skipping out on 20nm.

 

[Orangey]

The 14nm process Samsung gave to GF is a low power one, for phones etc.

I have not found any info on a high power process from GF below 20nm, not sure where their 20 is at either. It's really rather panic time for the GPU industry, past time for a new node and none are truly viable. All there is is the TSMC "16 finfet plus" which was up until now scheduled to enter volume production in 2016, with risk production late 2015.

Like a lot of TSMC processes it could end in tears and we might not get anything new until 2017 unless GF steps up.

 

[humbug]

The 14nm process Samsung gave to GF is a low power one, for phones etc.

I have not found any info on a high power process from GF below 20nm, not sure where their 20 is at either. It's really rather panic time for the GPU industry, past time for a new node and none are truly viable. All there is is the TSMC "16 finfet plus" which was up until now scheduled to enter volume production in 2016, with risk production late 2015.

Like a lot of TSMC processes it could end in tears and we might not get anything new until 2017 unless GF steps up.

Knowing GloFo thier high power process 20nm is over due already by 6 Months and will be another 6 months, and then another 6 month delay.... At the end of that it may be ready

 

[Rexton]

The 14nm process Samsung gave to GF is a low power one, for phones etc

So the 14nm process that AMD were meant to get from Samsung isn't being used for GCN?

http://wccftech.com/amd-tapes-20nm-14nm-finfet-chips-1h-2014/

What happened to AMD supposedly rolling out their first 20nm products in 1h 2014?

 

[Orangey]

The new 14nm-XM FinFET technology would be featured inside AMD’s low power product stacks which includes Mobile, Tablets and All-In-Ones.

20nm is not much better than 28, no idea what's going on with GF but one would assume if it was ready Apple would have snapped it all up. They're pretty much the only one who can afford the extremely marginal cost-benefit ratio.

 

[Rilot]

Apple A9? They're only on A7 at the moment with - I expect - A8 due in 12 months with the iPhone 6S; assuming they follow the same release schedule that they have for the last few years.
That would put the A9 at around 2017

New device -> Updated device with new SOC -> New device..... etc.

 

[Orangey]

TSMC claim they will have 10nm in 2017.

God knows how unless they learn a lot about new materials and fins in the next couple of years.

 

[StriderX]

I suspect Nvidia and AMD are getting rather irritated, wouldn't be surprised if they were slowly trying to replace TSMC at this point.

 

[Orangey]

Well Global Foundries used to be AMD's fabs, I think Nvidia has always been fabless. If they got together with someone like IBM or Samsung who have fabs perhaps they could have a whip round for the $20b or so it'd take to beat TSMC at this point...

Unfortunately we're going to have to endure a period of uncertainty as we barrel towards the end of the line for silicon. 2020 seems to be the point where Intel & TSMC will either be squeezing out the last drops on 5nm or using graphene, germanium, III-V or some other exotic substance.

http://en.wikipedia.org/wiki/List_of_semiconductor_materials
http://www.extremetech.com/computin...t-iii-v-cmos-wafers-roll-off-production-lines
http://www.extremetech.com/extreme/106899-beyond-22nm-applied-materials-the-unsung-silicon-hero
http://www.extremetech.com/computin...ds-if-you-ask-the-engineers-or-the-economists
https://www.semiwiki.com/forum/content/3502-chip-design-game-end-moore-s-law.html

 

[Kaapstad]

Edit