Rumor: TSMC preparing trial run of 10-nanometer ARM chips, could power future iPhones

2»

Comments

  • Reply 21 of 26
    bkkcanuckbkkcanuck Posts: 864member

    If this rumour and TMSC is almost ready with 10nm fabs this would be a huge hit for TMSC.....  

     

    Of course this would be important for Apple as well since having TMSC ready would mean that they would probably be able to get a larger portion of the fab production.  More powerful ARM based chips would be great.

  • Reply 22 of 26
    afrodriafrodri Posts: 190member
    Quote:

    Originally Posted by cali View Post





    YES!!!



    Someone on another article commented that the next iPhone will have everything needed and will plateau, saying after that there will be nothing more that can be added.



    Only people who lack imagination think we've reached the limit.

     

    While we certainly haven't hit the limit, we have reached a point of diminishing returns on the number of 'practical' transistors per chip and the impact this has on performance and power (i.e. the impact of moving form 90 to 65nm was far greater than 14 to 10nm will be). Already Moore's law seems to be slowing.

     

    For the practical limits of CMOS it is hard to say. Industry folks and the ITRS roadmap seem to get somewhat hazy after 5 or 7nm, though some might push it another generation. It also looks like there might be longer gaps between process node upgrades. There are a number of post-CMOS technologies which may or may not pan out.  From a pure physics perspective there is a lot of space to fit more computation in, though we will have to move from CMOS transistors to do so.

  • Reply 23 of 26
    ksecksec Posts: 1,569member
    Quote:

    Originally Posted by cali View Post





    I hope they nail Sammy's coffin.

     

    The reality is there are more customer willing to pay premium for leading node then the max capacity available.  Which means if Apple took most of those 10nm TSMC capacity, the remaining customer who cant get enough capacity from TSMC will have to move to Samsung anyway since there are no other choice.

     

    Intel has been quiet on Custom Foundry, I think the low ( or lower ) margin on Custom Foundry they are afraid of their EPS affecting stock price.

  • Reply 24 of 26
    wizard69wizard69 Posts: 13,377member
    cali wrote: »
    YES!!!

    Someone on another article commented that the next iPhone will have everything needed and will plateau, saying after that there will be nothing more that can be added.
    All technology plateaus.
    Only people who lack imagination think we've reached the limit.

    The iPhone will plateau when it is possible to do everything on something like an Apple Watch. Or maybe a lapel pin like on Star Trek. iPhone won't go no forever as a bleeding edge device.
  • Reply 25 of 26
    bkkcanuckbkkcanuck Posts: 864member
    Quote:

    Originally Posted by afrodri View Post

     

     

    While we certainly haven't hit the limit, we have reached a point of diminishing returns on the number of 'practical' transistors per chip and the impact this has on performance and power (i.e. the impact of moving form 90 to 65nm was far greater than 14 to 10nm will be). Already Moore's law seems to be slowing.

     

    For the practical limits of CMOS it is hard to say. Industry folks and the ITRS roadmap seem to get somewhat hazy after 5 or 7nm, though some might push it another generation. It also looks like there might be longer gaps between process node upgrades. There are a number of post-CMOS technologies which may or may not pan out.  From a pure physics perspective there is a lot of space to fit more computation in, though we will have to move from CMOS transistors to do so.


     

    Impact of size reduction of 90 to 65 and 14 to 10 is about the same... a 40% reduction in size.    

     

    As far as them being fuzzy after 7nm.... well they are always fuzzy when you are talking 2 to 3 generations down the road.... each reduction has risks.  They were trumpeting the death of Moore's Law 10 or 15 years ago .... I remember discussions about it even back then.  

     

    Personally, I think at a certain point the cycles will lengthen.... just because the increasing cost of each reduction in R/D and the risk .... the reduction in competition as competitors were knocked out of the industry and the fact that the volume / margin will eventually slow things down because of the lack of need of people to constantly replace something with something faster -- when they are not utilizing the performance of what they have....

     

    I will be happy when my phone has enough computing performance that I can use it as a phone or plug in a keyboard and external monitor and I can run "OS X" full edition on the monitor while still using iOS on the phone -- if someone calls.  The gains are all about mobile computing now.... 

  • Reply 26 of 26
    afrodriafrodri Posts: 190member
    Quote:

     

    Impact of size reduction of 90 to 65 and 14 to 10 is about the same... a 40% reduction in size.    

     


    You are correct that the size reduction is about the same, but the _impact_ of that size reduction is less. You almost never design a chip with all transistors at the minimum feature size. As we have moved to smaller process nodes the number of 'effective' transistors you can get and the potential power savings has been less than the minimum feature size would indicate. 

Sign In or Register to comment.