Ryzen vs i7 (Mainstream); Threadripper vs i9 (HEDT); X299 vs X399/TRX40; Xeon vs Epyc

Discussion in 'Hardware Components and Aftermarket Upgrades' started by ajc9988, Jun 7, 2017.

  1. ajc9988

    ajc9988 Death by a thousand paper cuts

    Reputations:
    1,597
    Messages:
    5,815
    Likes Received:
    8,206
    Trophy Points:
    681
  2. ole!!!

    ole!!! Notebook Prophet

    Reputations:
    2,385
    Messages:
    5,798
    Likes Received:
    3,716
    Trophy Points:
    431
    yea i'll believe the backport of 1st gen 7nm arch design to 10nm+++ cause that actually does sound reasonable. i mean we are at 10nm+ or ++ at the moment so it doesnt seem all that impossible. for the 7nm onward i have doubts but thats so far down the future no point talking about it.
     
  3. ajc9988

    ajc9988 Death by a thousand paper cuts

    Reputations:
    1,597
    Messages:
    5,815
    Likes Received:
    8,206
    Trophy Points:
    681
    That's cool, but I'm trying to get your timeline right. What you want is in 2022. That will likely be Zen 5 on 5nm+ with potentially a larger use of EUV. To put this into perspective, the 5nm non-plus will have 14 out of a total 59 possible layers done with EUV. It will have a transistor density around 173.1 MTx/mm^2. That is a 1.37x transistor density over Samsung and will be used by Apple, AMD, and HiSilicon in 2021. The plus variant is a refinement. I'm being AMD will use the refinement, waiting for 3nm to mature for 2023.

    That is double the density of transistors compared to Intel's 10nm node. That will allow AMD to go wider, which helps IPC.

    So even if Intel is using 10nm+++ at that time, they will have a process deficit in 2022. But, golden cove will have another large IPC gain as they are focusing on ST for that generation.

    The only question, and where we some down on different sides, is that you believe Intel will get to 5GHz, whereas I think Intel will be at 4.3-4.6 GHz. We can agree to disagree.

    But 10nm was allow AF and Intel killed it (also very low yields). 10nm+ is 900MHz-1.1GHz slower than 14nm+++. Intel was always over 4.4GHz on 14nm, seen in Broad well, with skylake hitting 4.8, 14nm+ kaby hitting 5.0, and now coffee refresh up to 5.3 if you can keep it cool. So it isn't out of the question. But Intel also had a large increase between broad well and skylake, whereas we have little to go on but it seems there wasn't much of any improvement in frequency from cannon to ice lake on frequency. So getting an additional GHz just doesn't make sense to me. Intel isn't magic, just like mortgages are not immune to the forces of the economy (which believing such gave us the 2008 recession).
     
  4. hmscott

    hmscott Notebook Nobel Laureate

    Reputations:
    7,072
    Messages:
    20,418
    Likes Received:
    25,218
    Trophy Points:
    931
    Where do you come up with these flights of fantasy? :)

    Zen 3 will be in the entire line up of Ryzen CPU's, from APU's to Epyc.

    Along the way Zen 3 will offer a big upgrade for EOL AM4's either then or down the road for owners of AM4 computers.

    I upgraded my AM2+ with the last best CPU available for that socket and it's still running fine 12+ years later.

    As for the Zen 3 ThreadRipper 3 and Epyc 3, who knows how much longer those sockets will continue or EOL as of today?

    One of the benefits of buying EOL motherboards and their last CPU's is the stability and long life they can provide. All the platform bugs are worked out, the prices have leveled off to reasonable prices, and if you do a good job on the build you've got a strong long term computer for many years of service.

    Buying the cutting edge all the time isn't a good idea for most people, as that's where the most problems and disappoints reside. Waiting for all of those bugs to get firmware updates gives you a stable platform to build on.

    In my personal life I prefer the stability and reliability of buying a platform that has been fixed and provides little or no distraction or drama.

    For work, that's a different story on one hand we have to keep at the cutting edge for competitive equality, but my penchant for stability and price vs performance focus is also of great value for companies that want the best bang for their bucks and low license, support, and service costs.

    It doesn't hurt to save as much money for the company as you cost them, on top of making money for them by keeping them competitive.

    Giving up on Zen 3 because it's EOL on AM4 is childish and not giving due consideration to building systems with stability and long life.
    This time you are invoking Intel's "fantasies", which likely will never happen just like all of Intel's 10nm fantasies have consistently failed to deliver in the end.

    Anything beyond that is even fuzzier speculation. Zen 4 may bring many new technologies, but with those new technologies will be a time of instability while the bugs get worked out.

    Reading your post again I gotta believe you really are lost in Intel's BS, when will you wake up and realize Intel is full of themselves and aren't delivering anything worth buying? - now or then, Intel's got a long way to go to successfully make it out of the woods.
     
    Last edited: Dec 12, 2019
  5. TANWare

    TANWare Just This Side of Senile, I think. Super Moderator

    Reputations:
    2,530
    Messages:
    9,535
    Likes Received:
    4,939
    Trophy Points:
    431
    Amd's hug problem is no one has full machine offerings right now. With the supply issues no OEM's are going to put out preconfigured machines. you need to be able to walk into B&M stores and have a good choice of machines. The same for business catalogs etc.. DIY's will never yield a significant overall market share.
     
    hmscott and ajc9988 like this.
  6. ole!!!

    ole!!! Notebook Prophet

    Reputations:
    2,385
    Messages:
    5,798
    Likes Received:
    3,716
    Trophy Points:
    431
    i am saying these on the assumption zen 5 will have similar IPC as golden cove. cause if it isn't theres no need to go AMD since intel has dropped price while offering higher clocks. zen3 will have about 15%? IPC over CFL if lucky so zen 4-5 hopefully will be another 15%. with IPC matched, obviously i will go with higher clocks even if i pay a bit more.
     
  7. rlk

    rlk Notebook Evangelist

    Reputations:
    132
    Messages:
    560
    Likes Received:
    292
    Trophy Points:
    76
    Talking about Zen 5 and Golden Cove right now strikes me as more than a bit speculative.
     
  8. ajc9988

    ajc9988 Death by a thousand paper cuts

    Reputations:
    1,597
    Messages:
    5,815
    Likes Received:
    8,206
    Trophy Points:
    681
    Shut it, because you look bad when you speak without hard data.



    And yes, zen 3 is over 15% IPC over coffee lake considering Zen 2 is already like 6-8% over it and zen 3 integer is 10% over that. Intel has lower IPC until ice lake.

    Further, look at those yields. 5nm over 80% yields. Intel can't get 10nm working and 7nm who knows what is happening.
     
    hmscott likes this.
  9. ajc9988

    ajc9988 Death by a thousand paper cuts

    Reputations:
    1,597
    Messages:
    5,815
    Likes Received:
    8,206
    Trophy Points:
    681
    https://www.anandtech.com/show/14605/the-and-ryzen-3700x-3900x-review-raising-the-bar/6

    Actual IPC results for Zen 2. Now stop talking without facts.
     
    hmscott likes this.
  10. ajc9988

    ajc9988 Death by a thousand paper cuts

    Reputations:
    1,597
    Messages:
    5,815
    Likes Received:
    8,206
    Trophy Points:
    681
    It is. He doesn't like AMD pulling ahead and has justified waiting until then, but is claiming that Intel will have what is scheduled for 2022 in 2021, then is comparing what should be available then to Zen 3 coming out 2020, then saying Intel wins. It is the most absurd thing I've heard in a long while.
     
    hmscott likes this.
Loading...

Share This Page