AMD Ryzen 5 2600 CPU Gets Delidded And Tested

Discussion in 'Frontpage news' started by Hilbert Hagedoorn, Apr 24, 2018.

  1. Hilbert Hagedoorn

    Hilbert Hagedoorn Don Vito Corleone Staff Member

    Messages:
    48,317
    Likes Received:
    18,405
    GPU:
    AMD | NVIDIA
    So before you go like huh wait .. didn't Ryzen have a soldered heat spreader on the die that cannot be removed? Yes you are right, but of course you can melt the Indium solder when it reaches 175 Deg...

    AMD Ryzen 5 2600 CPU Gets Delidded And Tested
     
  2. Srsbsns

    Srsbsns Member Guru

    Messages:
    192
    Likes Received:
    54
    GPU:
    RX Vega 64 Liquid
    I'm guessing this was done because his name is in Thermal Grizzly. Just a marketing opportunity.
     
  3. metagamer

    metagamer Ancient Guru

    Messages:
    2,596
    Likes Received:
    1,165
    GPU:
    Asus Dual 4070 OC
    It was the same with the first generation of Ryzen, can't remember if it was the 1800x or 1700x but delidding made next to no difference.
     
  4. reix2x

    reix2x Master Guru

    Messages:
    715
    Likes Received:
    246
    GPU:
    HIS 4870 1GB
    This was done because this is what Der8auer do for living, it could be marketing but what is the bad thing on marketing? this video pops adequate information, he even do not recommend to delid your ryzen 2, so i can't see the bad thing on this.
     
    Koniakki likes this.

  5. Yogi

    Yogi Master Guru

    Messages:
    351
    Likes Received:
    150
    GPU:
    Sapphire RX 6800
    Interesting about the die size. Would AMD get better yields if the architecture is more spread out?
    Obviously it would work out cheaper for them in other areas of production. They can utilize more existing tooling in the assembly process etc.
     
  6. __hollywood|meo

    __hollywood|meo Ancient Guru

    Messages:
    2,991
    Likes Received:
    139
    GPU:
    6700xt @2.7ghz
    id prefer direct die contact 10 times out of 10, but melting solder & thereby potentially damaging the core isnt worth four degrees ^^ good info, good video, good post
     
    schmidtbag and Embra like this.
  7. SaLaDiN666

    SaLaDiN666 Member

    Messages:
    32
    Likes Received:
    3
    GPU:
    1080 GTX
    Don't want to break it for you, but both Ryzens are 14nm.

    That "12nm" is just a marketing name for 14nm+.
     
    Yogi likes this.
  8. Srsbsns

    Srsbsns Member Guru

    Messages:
    192
    Likes Received:
    54
    GPU:
    RX Vega 64 Liquid
    Nobody said it was bad. I'm just saying its stunt for Thermal thermal grizzly that actually has his name on the package. Not a very useful one at that.

    He was so busy asking if he could. We never stopped to ask if we should. Now its answered.
     
  9. Koniakki

    Koniakki Guest

    Messages:
    2,843
    Likes Received:
    452
    GPU:
    ZOTAC GTX 1080Ti FE
    Saw the video earlier today. Good info, good watch.
     
  10. Yogi

    Yogi Master Guru

    Messages:
    351
    Likes Received:
    150
    GPU:
    Sapphire RX 6800
    That would make sense.
     

  11. fry178

    fry178 Ancient Guru

    Messages:
    2,067
    Likes Received:
    377
    GPU:
    Aorus 2080S WB
    Dont recommend it, but don't see a big risk in heating it up, unless you dont have something with a dial to set a temperature.
    The obviously didnt put cold solder on the cpu, so it survives the heat associated with soldering.
     
    Last edited: Apr 25, 2018
  12. Cooe

    Cooe Member

    Messages:
    39
    Likes Received:
    17
    GPU:
    AMD R9 Fury X 4GB
    This isn't true at all. It's a legit partial node shrink, with a 15% density improvement if used with the new 7.5T transistor library's it was released with (AMD chose to instead use the same 9T libraries as 14nm Ryzen 1 to increase dead-space between elements and maximize voltage efficiency improvements & thermal saturation reduction, instead of using the new 7.5T ones to actually shrink the die & increase yields. The latter of which would have required a full floor-plan redesign for a ≈15% increase in dies per wafer, but in exchange for far more work and a lesser effeciency/clock improvement. IMO they made the right choice, ESPECIALLY as it let the core Zen design team focus on the new 7nm floor-plan for Zen 2).

    Ryzen-G is using what's essentially GloFo's "14nm+", which like Intel's "+'s" has no real changes to element size but has been optimized from a long manufacturering period to reduce leakage. In fact, Intel's 14nm+ & ++ actually DECREASES the transistor density vs their bog standard 14nm as the means to make this leakage reduction happen. Intel's able to pull such a thing off because they don't fab their entire chips on the same node like AMD does, but rather only the cores & cache themselves on the leading node, with all the rest (uncore, iGPU, etc...) all fabbed on the prior process (22nm in this case), so small changes in the leading process' density/efficiency sweet-spot aren't a big issue if the clock-speed gains are worth it.
     
    Last edited: Apr 25, 2018
    __hollywood|meo likes this.
  13. Cooe

    Cooe Member

    Messages:
    39
    Likes Received:
    17
    GPU:
    AMD R9 Fury X 4GB
    Except that isn't actually true. GloFo's 12nm is a legit partial shrink with a 15% density improvement with the accompanying 7.5T transistor libraries. AMD's decision to use the same 9T library as Ryzen 1 and increase the amount of "dead-space" between elements instead, and thus keeping the die size the same as before was a deliberate choice made for very important reasons.

    https://www.anandtech.com/show/12625/amd-second-generation-ryzen-7-2700x-2700-ryzen-5-2600x-2600/2
     
  14. Cooe

    Cooe Member

    Messages:
    39
    Likes Received:
    17
    GPU:
    AMD R9 Fury X 4GB
    See my above 2 posts. Keeping the die size the same was an explicit choice of AMD's for very specific reasons and has nothing to do with GloFo's 12nm actually being a refined 14nm, aka "14nm+" with equal element size/possible density which just isn't true at all.

    It's a legit partial shrink (which was a much more common phenomena in the GPU world vs x86 prior to this), whereas Intel's "+" refinements actually go the opposite direction and decrease the transistor density as the means of reducing leakage & improving voltage characteristics to squeeze more out of a process w/o requiring major changes to element size; be in partial form like GloFo's 14nm to 12nm or TSMC's 16nm to 12nm; or an across the board shrink like Intel's 14nm to 10nm.

    Intel can get away with this kind of "optimization" because as foundry owners they're able to, and thus always have fabbed only the cores & cache on their leading node, meaning a slight reduction in it's density won't have a major effect on overall yields, as it's counteracted by the process' increasing maturity. This style of optimization can only be pushed so far though as the reduced density has to be counteracted by increased yields, meaning it's ofc only feasible on a mature process and can only be pushed to a certain extent. Coffee Lake's 14nm++ is likely the end of that line for their 14nm process.
     
    Last edited: Apr 25, 2018
  15. SaLaDiN666

    SaLaDiN666 Member

    Messages:
    32
    Likes Received:
    3
    GPU:
    1080 GTX
    Overall, GlobalFoundries has stated that its 12LP process offers a 10% performance improvement and a 15% circuit density improvement over 14LPP.

    https://i.imgur.com/GiCYddm.png

    It is compared to 16nm of TSMC.

    Not between 14nm and 12nm of GLO.
     

  16. RzrTrek

    RzrTrek Guest

    Messages:
    2,548
    Likes Received:
    741
    GPU:
    -
    I would think a decrease of 4 degrees is good, but not enough to void your warranty.
     
  17. metagamer

    metagamer Ancient Guru

    Messages:
    2,596
    Likes Received:
    1,165
    GPU:
    Asus Dual 4070 OC
    It's a fair result, I guess. The temp difference will not be 4 degrees across the board also, at 80C it'll be different and at 50C it'll be different. I wouldn't myself delid because it's not really worth it. That being said, my i7 is delidded but the difference there is large because of the TIM Intel use.
     
  18. Amx85

    Amx85 Master Guru

    Messages:
    335
    Likes Received:
    10
    GPU:
    MSI R7-260X2GD5/OC
    no that RYZEN CPU auto-overclocks deppending on temps? he must show performance results @stock clocks, just a stock temp test does´nt looks cool and less, just showing 4º difference overclocked

    maybe i delid my A6-9500 after buy the R5 lol

    greetings
     

Share This Page