Ram difference - Ryzen 7 1700 with Samsung B-Die

Discussion in 'AMD Processors' started by Xa3phod, Jan 14, 2019.

  1. Xa3phod

    Xa3phod Limp Gawd

    Messages:
    137
    Joined:
    Sep 19, 2012
    Just wondering. What is the performance difference for a Ryzen 7 with Samsung B-Die and something a bit slower? I know Ryzen loves faster ram. If I have DDR4-3200, but its CL16-18-18-38. How about CL14-14-14-32? Would you see a big impact on performance?

    Thanks!
     
  2. Wormfeeder

    Wormfeeder n00b

    Messages:
    12
    Joined:
    Nov 18, 2018
    I have DDR-3200 @ 16-18-18 cinebench score is around 1300. I set the ram to 14-16-16 and cinebench runs at 1340ish. so yeah timings make a difference.
     
  3. hititnquitit

    hititnquitit Gawd

    Messages:
    515
    Joined:
    Apr 16, 2018
    if your budget allows for a $200 2x8gb set of b-die modules go for those. if not, the difference between c14 and c16-18 is minimal at best and cost a heck of alot less.
     
    kirbyrj and Xa3phod like this.
  4. Xa3phod

    Xa3phod Limp Gawd

    Messages:
    137
    Joined:
    Sep 19, 2012
    What % performance difference between DDR4 3200 and DDR4 3000 with a Ryzen 7 1700 overclocked to 3.9Ghz? Any noticeable difference?
     
  5. Nobu

    Nobu 2[H]4U

    Messages:
    3,234
    Joined:
    Jun 7, 2007
    I can't answer this, exactly, as my 1600 wont quite reach 3.9 (it struggles with 3.75+) and I'm not sure how stable my ram would be at 3200, but I can give it a shot at 3.75 and let you know how that goes. I have micron hynix mfr (drunk/sick/tired) so it won't be 1:1 with samsung, but it'll be similar enough if I can hit 3200 stable.
     
    Last edited: Jan 15, 2019
  6. N4CR

    N4CR 2[H]4U

    Messages:
    3,863
    Joined:
    Oct 17, 2011
    Someone on here was getting Cl14 with cheap ripjaws V 3600 dual rank.. (which is stock at CL19). If you're [H] this is the way to do it on the cheap, I replied to their thread and I think it's findable with CL14 ripjaws or similar.
     
    kirbyrj likes this.
  7. Nobu

    Nobu 2[H]4U

    Messages:
    3,234
    Joined:
    Jun 7, 2007
    So here we go:
    3000c14-15-14-28
    19.21GB/s theoretical
    1216 cb r15 multithread
    150 cb r15 single thread

    3200c16-17-17-30
    19.16GB/s theoretical
    1185 cb r15 multithread
    150 cb r15 single thread

    Assuming you get sticks with similar timings to mine, it'll be more or less a wash. If you get 3200 with better timings, you should see better performance than mine at 3000. If you get 3400/3600 with similar timings to mine at 3200, you should also see better performance (likely much better, if you can get it stable).

    I could see a definite improvement in cb, but it was only a small bump mostly in multithread. In games it may be more or less obvious, depending on the game. I used the fast settings from dram calculator for both 3000 and 3200.
    I have G.Skill Ripjaws V F4-2800c16d-16gvg (hynix "m-die" mfr, single rank), ymmv.
     

    Attached Files:

    c3k and Xa3phod like this.
  8. ccityinstaller

    ccityinstaller [H]ardness Supreme

    Messages:
    4,229
    Joined:
    Feb 23, 2007
    If you are gaming, or doing almost anything other then playing benchmark extreme! then just buy the cheaper kit and see if you can tighten the timings a bit. I am using a C16-18-18 kit that I have tightened up @ 3200 and do not notice a difference in anything except CB scores. Mining, games, browsing etc is all the same.

    I would take that $150~200 you save and put toward a faster GPU (if you are in the market for one) or buy a speedy 1TB MVNE SSD for plenty of game storage and keep the difference.
     
  9. trick0502

    trick0502 [H]ardness Supreme

    Messages:
    5,154
    Joined:
    Apr 17, 2006
    I’ll do some test on my 2700 tonight.I have Corsair rgb 3200 with some bad timings.
     
    alxlwson likes this.
  10. bobzdar

    bobzdar [H]ard|Gawd

    Messages:
    1,645
    Joined:
    Jun 6, 2003
    3%.
     
  11. Joust

    Joust 2[H]4U

    Messages:
    2,933
    Joined:
    Nov 30, 2017
    Using Ripjaws V 3200 (16gb x 2) on a 1950x - can't get it stable at 3200. Frankly, I cannot justify the B-die price, though maybe I'll go that route sometime in the future if prices continue to decrease.

    ...And that price-value is coming from a Threadripper user. That said, I'm not very [H]ard at all.
     
  12. LigTasm

    LigTasm [H]ardness Supreme

    Messages:
    4,977
    Joined:
    Jul 29, 2011
    I have a set of Corsair 3200 C15 that performs nearly identically to my Gskill 3200 C14 B-dies at about $70 cheaper. For gaming there is zero difference, less than 5% for synthetics.
     
  13. Furious_Styles

    Furious_Styles [H]ard|Gawd

    Messages:
    1,383
    Joined:
    Jan 16, 2013
    Yeah it's very marginal and game specific. It's like going fro 4.9->5.0 ghz. Sure there's a gain but you won't notice it.
     
  14. OnceSetThisCannotChange

    OnceSetThisCannotChange Limp Gawd

    Messages:
    142
    Joined:
    Sep 15, 2017


    It can be up to 25%, depends on usage.
     
  15. Lunas

    Lunas [H]ardForum Junkie

    Messages:
    9,852
    Joined:
    Jul 22, 2001
    the difference all comes down to compatibility and timings. non b die ram had compatibility issues with ryzen it would not run at speed b die ram had the best compatibility with speeds closest to rated speeds and timings.
    i have some non b-die ram 3200 mhz rated only runs at 2933mhz.

    Ryzen likes fast ram but it is picky about the ram it will run.
     
  16. Furious_Styles

    Furious_Styles [H]ard|Gawd

    Messages:
    1,383
    Joined:
    Jan 16, 2013
    OP is talking about going from 3200mhz C16 -> 3200mhz C14.
     
  17. OnceSetThisCannotChange

    OnceSetThisCannotChange Limp Gawd

    Messages:
    142
    Joined:
    Sep 15, 2017
    B die will enable you to push further, and if you tweak it to the max you can get significant gains.

    Project Cars 2 is that outlier above with 25% difference between 3200 with regular timings and B die pushed to the best timings at 3333, as it is memory sensitive. The average across all games was around 10%.
     
  18. gerardfraser

    gerardfraser Gawd

    Messages:
    911
    Joined:
    Feb 23, 2009
    I have tested many combinations of ram in a few games gaming 2560 x 1440 ultra.This is what I got and no I did not put every test I did in video or chart or test every game in the world LOL.

    CL14 2933Mhz - CL14 3600Mhz you get the same FPS in gaming and that is the sweet spot.

    CL16 3200Mhz to CL14 3466MHZ % breakdown and for the OP going from say CL16 and buying CL14 is really not worth the trouble.

    xtum8RM.png
    Desktop-Screenshot-2018-12-08-11-11-27-03.png

     
    Last edited: Jan 18, 2019
    Darkstalker420, c3k and Nobu like this.
  19. Astral Abyss

    Astral Abyss 2[H]4U

    Messages:
    2,660
    Joined:
    Jun 15, 2004
    For me, the benefit of running B-die is that I was able to populate all 4 sockets for 32GB and still keep 14-14-14-34-1T timings. Good luck doing that with other RAM.
     
    noko likes this.