The AnandTech Podcast, Episode 46: More Choice on x86

22 Comments

View All Comments

• ಬುಲ್ವಿಂಕಲ್ ಜೆ ಮೂಸ್ - Wednesday, May 9, 2018 - link

  Would any of you prefer an 1st gen 10nm INTEL CPU to a highly optimized 14nm Whiskey Lake, assuming they both have the same Meltdown Fixes
  
  I have been holding out for 10nm but am beginning to think a highly optimized process is the way to go at this point

• ಬುಲ್ವಿಂಕಲ್ ಜೆ ಮೂಸ್ - Wednesday, May 9, 2018 - link

  Lets say Whiskey Lake came out this evening and Canon Lake came out tomorrow morning
  
  They both are identical as to cores/clocks and cost
  
  Would you prefer a 1st gen/revision-0 10nm process just to save 5-10 power consumption or a highly optimized 14nm?

• GreenReaper - Thursday, May 10, 2018 - link

  First gen 10nm, because it's likely to be the optimized stuff but shrunk down, and that power reduction possibly means higher frequency as well. But I wouldn't buy anything the day of release, I'd wait for the reviews,

• Krysto - Thursday, May 10, 2018 - link

  Neither. I think it would be a mistake to buy any of these scramblings from failure from Intel at this point. This happened with Broadwell do. Intel delayed it a for a long time, and it ended-up one of the worst microarchitectures in since Sandy Bridge.

• Dragonstongue - Thursday, May 10, 2018 - link

  ^^^^ time will tell, but they are very much in scramble mode right now, maybe they will "get it right" but then again, maybe they will not.
  
  in theory the jump down to 10nm will "help" but then again it is very "unknown" at this point so any possible problems or gains are likely to surface on gen 1 of the 10nm shrink....whereas a "new" chip on a very very known 14nm process will likely get all the possible love they can give it.
  
  I think something along the line of Ryzen gen 1 (Ryzen 1xxx) on 14nm are quite good, but the refinement of this 14nm down to "12nm" for Ryzen gen 2 (2xxx 2018) helpe in regards to more constant more aggressive turbo core and/or reduction in power at the same clocks.
  
  usually a smaller node allows the reduction in power sometimes an increase in clockspeed for less power or increase in clock for the same power at least it has more or less been this way for many generations for AMD as well as Intel more Intel, but they have been more about doing minimal performance +% vs reduction in power -%.
  
  It has to get them grumbling though that all their fab competition (TSMC, GF, Samsung etc) will all be hitting 7nm likely within a scant few months prior to or slightly after Intel finally manages to get to 10nm.
  
  well unless they all had some kind of wonky agreement to do exactly this (unlikely, but sometimes even the worst of "enemies" can see eye to eye in fighting a war)

• Megatomic - Wednesday, May 9, 2018 - link

  There is a problem with this podcast episode in Pocket Casts and it cannot be downloaded. Can you please look into this?

• Ian Cutress - Wednesday, May 9, 2018 - link

  It should be sorted now - can someone confirm for me please :)

• Megatomic - Thursday, May 10, 2018 - link

  Ian, I've tried on two different devices now. OP3T on stock OOS 5 with the latest Pocket Cast app and also my P2XL on DP2 with the latest Pocket Cast app. OP3T on AT&T and P2XL on VZW. I've tried both phones on WiFi and LTE at home and at work. Neither will download this episode. I'll just listen to the podcast using the weblink you gave me on Twitter, thanks for checking on this. Hopefully episode 47 doesn't have this problem.

• Megatomic - Monday, May 14, 2018 - link

  Ian, just wanted to let you know that I fixed the problem I was having. I unsubscribed and then subscribed again, I was able to download the episode after that. So weird but anyways that was it.

• tipoo - Thursday, May 10, 2018 - link

  71:52 FIN(fet)