[KitGuru] TSMC on track för 10 nm @ end of 2016

[Mondozei]

What do you think?

 

[Elixer]

http://www.digitimes.com/news/a20150717PD208.html

TSMC has reiterated plans to move 10nm FinFET technology to risk production at the end of 2016, followed by mass production in the first quarter of 2017.

Commercial production of chips made on the new process will take place in TSMC's manufacturing site at the Central Taiwan Science Park (CTSP) in Taichung, the foundry disclosed.

Intel recently confirmed that its 10nm process will not be ready for volume production until the second half of 2017. TSMC might beat Intel in the 10nm race.

http://appleinsider.com/articles/15...0nm-chips-by-early-2017-on-pace-to-beat-intel

The race to make smaller and more efficient mobile processors continues, with iPhone chipmaker Taiwan Semiconductor Manufacturing Co. revealing it remains on track to mass produce its first 10-nanometer FinFET processors by early 2017 — a timeframe that would put it ahead of rival Intel.

 

[Silverforce11]

What they plan to achieve is different to what they can or will achieve.

If they do beat Intel, it would be an amazing miracle.

 

[ShintaiDK]

What they plan to achieve is different to what they can or will achieve.

If they do beat Intel, it would be an amazing miracle.

You do know 16FF is just 20nm with FF right? So where do you think that places "10nm"?

 

[HurleyBird]

You do know 16FF is just 20nm with FF right? So where do you think that places "10nm"?

Was going to say the same thing. Is it 10nm or "10nm"? And what are the chances TSMC's predictions aren't total bunk yet again?

 

[pepone1234]

So in 2017 TSMC will finaly be on par with intel 14nm?

 

[ShintaiDK]

So in 2017 TSMC will finaly be on par with intel 14nm?

Something like that with an LP process with low yields. TSMC never delivered on time before. Usually 2 years more.

TSMC and on track doesnt mix.

 

[Samwell]

So in 2017 TSMC will finaly be on par with intel 14nm?

If they bring it on time, which i don't believe, then Intels process advantage would be nearly gone. Density will be way more comparable to Intel 10nm than 14nm.

 

[shady28]

Something like that with an LP process with low yields. TSMC never delivered on time before. Usually 2 years more.

TSMC and on track doesnt mix.

Well, Intel and on track doesn't mix either, at least not for the last 2 years. This is just status quo, they all fell behind so Intel maintains a 1+ year lead.

Except for Samsung. That was the dark horse that came out of nowhere.

 

[ShintaiDK]

Well, Intel and on track doesn't mix either, at least not for the last 2 years. This is just status quo, they all fell behind so Intel maintains a 1+ year lead.

Except for Samsung. That was the dark horse that came out of nowhere.

I fully agree there.

 

[IntelUser2000]

Well, Intel and on track doesn't mix either, at least not for the last 2 years. This is just status quo, they all fell behind so Intel maintains a 1+ year lead.

Except for Samsung. That was the dark horse that came out of nowhere.

2001-0.13u, Mid-2001 product
2003-90nm, Early-2004 product
2005-65nm, Late-2005 product
2007-45nm, Late-2007 product
2009-32nm, Early-2010 product
2011-22nm, Early-2012 product
2013-14nm, Late 2014 product

Initially the dates on their graph roughly coincided with actual product launch. But now the dates are "production". They make up dates just as the rest do.

 

[GreenOrbs]

I believe that Intel has announced that products with the 10nm node will be delayed until the second half of 2017. Apparently they are moving to 2.5 year cycles vs. the previous 2 year cycle.

http://www.businessinsider.com/intel-saves-1-billion-moores-law-2015-7

I think it will be interesting if the other players manage to catch up with Intel. Looks like there might be a possibility of that happening.

 

[3DVagabond]

I believe that Intel has announced that products with the 10nm node will be delayed until the second half of 2017. Apparently they are moving to 2.5 year cycles vs. the previous 2 year cycle.

http://www.businessinsider.com/intel-saves-1-billion-moores-law-2015-7

I think it will be interesting if the other players manage to catch up with Intel. Looks like there might be a possibility of that happening.

Considering the delay and trouble for 20nm from TSMC I'm not banking on them "catching up". More likely that they are being overly optimistic.

 

[alcoholbob]

You do know 16FF is just 20nm with FF right? So where do you think that places "10nm"?

I've heard that 10nm was in fact going to be a 1.5 node shrink from 16FF.

20nm silicon -> 10nm silicon (double node shrink)
16nm FinFets -> 10nm FinFets (single shrink)

(2 + 1) ÷ 2 = 1.5 nodes

Now how TSMC goes from 20nm silicon to beating Intel to 10nm might require some space magic.

 

[JDG1980]

I believe that Intel has announced that products with the 10nm node will be delayed until the second half of 2017. Apparently they are moving to 2.5 year cycles vs. the previous 2 year cycle.

http://www.businessinsider.com/intel-saves-1-billion-moores-law-2015-7

I think it will be interesting if the other players manage to catch up with Intel. Looks like there might be a possibility of that happening.

Samsung might have a chance of catching up with Intel. TSMC? Probably not.

 

[Borealis7]

what do i think? i think we're going to have some pretty powerful smartphones in the coming years

and that makes me sad.

 

[ClockHound]

what do i think? i think we're going to have some pretty powerful smartphones in the coming years

and that makes me sad.

LOL!

No. Wait. That makes me sad too.

We will all woe the day EK is reduced to making designer liquid cooling blocks, er.. cases for the iPhone 10.

 

[Mondozei]

Wouldn't be surprised if Samsung got ahead to 10 nm before everyone else.

 

[beginner99]

Wouldn't be surprised if Samsung got ahead to 10 nm before everyone else.

Still processes don't have to be comparable and AFAIK Intel is the only one that got a high performance process below 28 nm, actually 2 of them 22 and 14 nm. So a proccess for small, low power SOCs isn't really comparable to a process usable for 600mm^2 Server CPUs.

 

[Stuka87]

Even if they do get 10nm out "on time", its unlikely that it will be for HP chips. It will most likely be low power SoCs and the like. Just like with 20nm, HP solutions delayed so long it was just skipped.

 

[RaulF]

Wouldn't be surprised if Samsung got ahead to 10 nm before everyone else.

This.

TSMC, not so much.

 

[badb0y]

So what we are saying is 2018 confirmed?

 

[Pwndenburg]

Whomever finds the next solution aside from silicon should be awarded the nobel peace prize and an angry joe Bad*** seal of approval.

 

[Sabrewings]

Whomever finds the next solution aside from silicon should be awarded the nobel peace prize and an angry joe Bad*** seal of approval.

IBM used a new silicon alloy for their 7nm chip, but I'm not sure how much further that will scale. It might be a one trick pony and won't work past 7nm.

 

[Pwndenburg]

Well the only thing I can say for certain is that if it is fixable I won't be the one to do it But I really do hope there is something beyond silicon that will one day become viable for a whole new "computer revolution."