• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

TSMC Aiming for 10nm by 2016

That's a fairly large jump in lithography process size from 2015-2016 if true, where does 16nm fit into this picture? I can see truth in Greg's comment to be honest: 20nm 2015, 16nm 2016-17, 10nm 2018. And beyond that? Well, we are getting dangerously close to the atomic size limit - within the next few years I suspect that research and development on "non-conventional" (conventional currently being silicon transistor based technology) systems will ramp up, as we do need an alternative for the future. Quantum computing could be the alternative we require, but for such a technology to reach mainstream markets will still take about 20 years - so we need an intermediary!
 
PhysicsMan94 said:
That's a fairly large jump in lithography process size from 2015-2016 if true, where does 16nm fit into this picture? I can see truth in Greg's comment to be honest: 20nm 2015, 16nm 2016-17, 10nm 2018. And beyond that? Well, we are getting dangerously close to the atomic size limit - within the next few years I suspect that research and development on "non-conventional" (conventional currently being silicon transistor based technology) systems will ramp up, as we do need an alternative for the future. Quantum computing could be the alternative we require, but for such a technology to reach mainstream markets will still take about 20 years - so we need an intermediary!
Click to expand...

Here's some more industry info:

http://www.cadence.com/Community/bl...-finfet-plus-10nm-and-7nm.aspx?postID=1333670

https://www.semiwiki.com/forum/content/3427-tsmc-updates-20nm-16nm-10nm.html

I'm skeptical as well, but mostly about 10nm. It sucks that the only competent fabs don't make high-end GPUs. :mad: :mad: :D
 
Any word on whether they have managed to bring the price per wafer down to an economically sustainable level? I seem to remember Nvidia hammering them for the price of their 28nm process, and the problem was only going to get worse as shrinkage continued.

Perhaps 10nm will see the switch to 450mm wafers?
 
Orangey said:
http://wccftech.com/tsmc-aiming-10nm-node-2016-20nm-production-increased-q3/
Click to expand...

I have a feeling that one or both are switching to GF for 20nm because (as far as i understand it) GF hold a GDDR stacking production licence.

I don't know if TSMC have acquired one of thier own but they have also upset AMD and Nvidia by jacking the price of 20nm right up and then give priority to Apple.

Perhaps TSMC have made some changes to win them back? there was defiantly talk of switching to GF.
 
Someone on here has been swearing blind that AMD are dropping TSMC completely after 28nm, it seems Nvidia are stuck with it though for Maxwell. Unless they go back to the drawing board a 2nd time...
 
humbug said:
I have a feeling that one or both are switching to GF for 20nm because (as far as i understand it) GF hold a GDDR stacking production licence.

I don't know if TSMC have acquired one of thier own but they have also upset AMD and Nvidia by jacking the price of 20nm right up and then give priority to Apple.

Perhaps TSMC have made some changes to win them back? there was defiantly talk of switching to GF.
Click to expand...

While in real terms TSMC did jack up the prices, they only did so because price to produce has increased. It's all estimations but, a wafer takes weeks to go through a fab from start to finish, double patterning is being introduced by Intel and TSMC for this next round of processes(14nm for Intel but with 64nm metal pitch, 20-16nm for TSMC/GloFo, also 64nm metal pitch). This will increase the amount of stages, the amount of time to manufacturer. If a fab can process say 50k wafers a month then you double the length of the process, it will now be 50k wafers every two months. AS things get more complex costs increase significantly.

It's happened multiple times before, we pass though(without remembering the stage precisely) say 180nm to 150nm and wafer cost increases dramatically as they need something much more complex, say immersion lithography. SO you switch from 150mm wafers to 225mm wafers. While you get less wafers per month, you can get more chips per wafer. The bigger wafers are used to offset increasing complexity and keep the chip output relatively speaking on track over time.

They were due 450mm for 20nm but it feel behind and got pushed back by everyone in the industry. So 14-10nm(real, not marketing, so sub 64nm metal pitch), I've not read what will happen to costs, 450mm should bring costs down but moving to EUV(assuming it's still the plan) is likely to increase costs. Probably at worst no big increase in costs from 20 to 10nm.

They might be playing with 10nm, in fact should be, by 2016 but they were playing with 20nm production in early 2013 iirc. Tape outs are taking longer, ramps are taking longer, cost are increasing for a tape out, doubt we'd see a 10nm gpu before 2017. 20nm gpus in 2015, 16nm gpus in 2016, 10nm in late 2017/2018 maybe?
 
They can keep 10nm, I will settle for 20nm a bit quick.

Where is this news from, wccftech - forget it just a third rate rumour.
 
You must log in or register to reply here.
Back
Top Bottom