1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Intel kills 10nm ?? oO

Discussion in 'CPUs' started by Zeed, Oct 22, 2018.

  1. kleox64

    Gangster

    Joined: Jul 28, 2005

    Posts: 399

    150c with 10 year minimum life, with little to no degradation in function and performance isn’t willy nilly...
     
  2. 4K8KW10

    Soldato

    Joined: Sep 2, 2017

    Posts: 6,946

  3. Journey

    Soldato

    Joined: Oct 18, 2002

    Posts: 6,303

    Location: West Midlands

    Pretty sure that the giveaway with 10nm not happening soon/at all was the almost giveaway prices on the 10xxx x299 CPU's. They've had to bite the bullet, and cut pricing due to the inability to compete on the desktop in cores/threads and TCO/PPW.
     
  4. Rroff

    Man of Honour

    Joined: Oct 13, 2006

    Posts: 64,304

  5. Robert896r1

    Hitman

    Joined: Sep 28, 2018

    Posts: 538

  6. NinjaCool

    Hitman

    Joined: Mar 27, 2010

    Posts: 517

    Location: Denmark

    It will be a big win for AMD if this turns out to be true :eek:
     
  7. 4K8KW10

    Soldato

    Joined: Sep 2, 2017

    Posts: 6,946

    Great time for AMD to pay the $1B debt back, since no competition in the foreseeable future :)
    I think they need to rethink their ongoing projects spendings.
     
  8. Robert896r1

    Hitman

    Joined: Sep 28, 2018

    Posts: 538

    They had to spend to get there they are now or bleed out slowly. It was clearly worth it.
     
  9. kleox64

    Gangster

    Joined: Jul 28, 2005

    Posts: 399

    Intel reps don’t have up to date information, 10nm is shipping with HEDP/Server parts in validation... this was still the case 6 months ago.

    It’s the validation causing the issues.
     
  10. kleox64

    Gangster

    Joined: Jul 28, 2005

    Posts: 399

    Also we have no reference to what 7nm means for Intel, again marketing fudging the numbers exactly what the other foundries are doing... this is common practice unfortunately.
     
  11. Orange Nexus

    Hitman

    Joined: Sep 21, 2018

    Posts: 710

  12. StriderX

    Capodecina

    Joined: Mar 18, 2008

    Posts: 21,539

    Clearly on purpose, "LOOK AT US, WE'RE STILL DOING THINGS".
     
  13. Aretak

    Wise Guy

    Joined: May 26, 2014

    Posts: 1,769

    We'll probably end up with them shuffling out a couple of underwhelming SKUs in a couple of years like they did with Broadwell, just so they can say that they technically released 10nm chips on desktop.
     
  14. Boot_Loop

    Gangster

    Joined: Jul 12, 2016

    Posts: 186

    Still, they're aiming for 10nm desktop in 2H 2021. That'll be going up against Ryzen 5000 on TSMC's 5nm process node. They are going to be in a difficult spot for years.

    I predict AMD going to gain serious marketshare in desktop and about 25%+ in server by 2021.
     
  15. alec

    Gangster

    Joined: Dec 11, 2016

    Posts: 329

    never mind the nanometers.When will desktop see magical IPC increase from Sunny Cove?
     
  16. chrcoluk

    Mobster

    Joined: Feb 27, 2015

    Posts: 3,720

    They will if it has the right performance metrics.

    Would you buy a 14nm chip faster and better priced than a competitive 7nm chip? Or would you buy just because 7nm is a smaller manufacturing process?
     
  17. 4K8KW10

    Soldato

    Joined: Sep 2, 2017

    Posts: 6,946

    Let's repeat it again:

    Great time for AMD to pay the $1B debt back, since no competition in the foreseeable future :)
    I think they need to rethink their ongoing CPU projects spendings and redirect more towards GPU segment and debt repayment.

    The thing about debt is that it's better to have none, so in the future they could ask for some if needed.

    I think they pay it back quite slowly.

    There are multiple articles on the matter and 5nm TSMC = 7nm intel.

    lol i7-5775C was the best i7 ever made :D
     
  18. DragonQ

    Soldato

    Joined: Jun 13, 2009

    Posts: 6,709

    Broadwell-C should've been a bigger clue to most of us of what was to come: no more die shrink easy wins or tick-tock. It was fine with Ivy Bridge because stock clocks weren't maxed but even then people cottoned onto the fact that the older 32nm Sandy Bridge clocked higher. In theory a better model would actually be to shrink mobile chips first (for efficiency gains) and then have desktop parts follow once the process is more mature and can reach higher clock speeds, which is what Intel seems to be shifting to now.

    AMD are doing things the opposite way around, which is probably not sustainable. It worked with Zen 2 because their previous GF nodes were just a bit pants, but if 5nm can't clock as high as 7nm+ they're going to need another 10-15% IPC jump to continue with their current release cadence.
     
  19. sidthesexist

    Hitman

    Joined: Jul 23, 2006

    Posts: 899

    Even if Intel move to a smaller node how are they going to overcome the thermal issues, Given they will have a higher transistor density and as far as we know still be using a monolithic core rather than a chiplet style design? Wouldn't this make their thermal issues worse than they are now?
     
  20. Boot_Loop

    Gangster

    Joined: Jul 12, 2016

    Posts: 186

    The peoblem with that is 14nm has barely no room to improve whereas 7nm to 7nm+ next year (Ryzen 4000) is going to see another comparatively huge leap forward.

    Already, compare Ryzen 8-core 3700X and Intel 9700K equivalent. Very similar performance in games yet the Ryzen draws so much less power while doing it, despite 8 more threads. Also, the 3700X wipes the floor with that in multi-threaded perf on top.