MathBenchmark comparison thread

Discussion in 'Overclocking & Hardware' started by ginger_nuts, Oct 25, 2020.

  1. OP
    OP
    ginger_nuts

    ginger_nuts Member

    Joined:
    Sep 24, 2011
    Messages:
    1,314
    Location:
    Morphett Vale, SA
    That an older chip is getting a similar score to a newer one.


    Not my bench, Shane is the self-proclaimed master at this. I am just trying to wrap my head around it.


    [​IMG]
     
  2. spit051261

    spit051261 Member

    Joined:
    Sep 29, 2019
    Messages:
    2,810
    Location:
    Perth
    Ill try a 10980xe for comparison.
    Is this a bench that prefers RAM over CPU ?
     
  3. OP
    OP
    ginger_nuts

    ginger_nuts Member

    Joined:
    Sep 24, 2011
    Messages:
    1,314
    Location:
    Morphett Vale, SA
    I once got told that you can never replace CPU speed with memory speed.

    But not sure when it comes to this bench.

    I hope in the next day or two to get my old gear out
     
  4. dasa2

    dasa2 Member

    Joined:
    Feb 5, 2003
    Messages:
    841
    Location:
    Northern VIC
    I actually scored higher with my RAM at 2133 SPD than at 3600c14 tweaked.
    There was some background processes that would account for the drop but I don't think RAM makes a big difference in this test.

    It was using 100% of my 7700K so I would say more cores do matter but then my score is only ~900 while the slower 7350K above scored almost double. Is that with something other than win10?

    Edit ok thanks to ginger_nuts for suggesting I disable HT.
    So I played around with Different Affinity HT on vs OFF and safe to say just disable HT and forget about it.

    So after booting into windows 10 diagnostic startup to take care of the inconsistency created by my background apps and disabling HT.

    7700K@4.7GHz core & cache 3600c15 tweaked subs Score: 1513
    7700K@4.7GHz core 4.2GHz cache 2133SPD Score: 1512
     
    Last edited: Oct 27, 2020
    ginger_nuts likes this.
  5. MR CHILLED

    MR CHILLED D'oh!

    Joined:
    Jan 2, 2002
    Messages:
    155,542
    Location:
    Omicron Persei 8
    More benchies :leet: I'll give this a go shortly :thumbup:

    EDIT: Intel 10900K @ 5GHz, PC3200

    Bench.JPG
     
    Last edited: Oct 27, 2020
  6. shane41

    shane41 Member

    Joined:
    Nov 10, 2008
    Messages:
    6,824
    Location:
    dice que no sabe
    New Intels all same similar. Anything from say 7th gen to 10th gen. Yes 1564 looks right :thumbup:
     
  7. shane41

    shane41 Member

    Joined:
    Nov 10, 2008
    Messages:
    6,824
    Location:
    dice que no sabe
    Bench does like high speed mems. Keep going with above idea ;)
     
  8. shane41

    shane41 Member

    Joined:
    Nov 10, 2008
    Messages:
    6,824
    Location:
    dice que no sabe
    Hurt score. See test 1. You have to back something off OR not stable.
     
  9. MR CHILLED

    MR CHILLED D'oh!

    Joined:
    Jan 2, 2002
    Messages:
    155,542
    Location:
    Omicron Persei 8
    It's stable. Just running 3200 CL15 ram. No tweaks other than the cpu at 5GHz. So this benchmark loves mem tweaks yeah?
     
  10. dasa2

    dasa2 Member

    Joined:
    Feb 5, 2003
    Messages:
    841
    Location:
    Northern VIC
    shane41 seems to think so but it makes no difference at all on my system so it may depend on the systems cache setup not sure.
     
  11. MR CHILLED

    MR CHILLED D'oh!

    Joined:
    Jan 2, 2002
    Messages:
    155,542
    Location:
    Omicron Persei 8
    I have some PC3600 CL16 coming soon so I'll try that ram with the same system, and not changing anything else.
     
  12. dasa2

    dasa2 Member

    Joined:
    Feb 5, 2003
    Messages:
    841
    Location:
    Northern VIC
    Ok here we have
    7700K@5.1GHz core 5GHz Cache, 3600c14 =1642
    7700K@5.1GHz core 5GHz Cache, 3200c14 =1642
    7700K@5.2GHz core 4.2GHz Cache, -1 AVX, 2400c15 =1673
    Last one has a 1.96% increase in clock speed for 1.89% increase in score.

    Also completed a test with 5.1GHz cache but it was unstable and as such it also booted with lose RTL resulting in 39ns RAM latency vs 36.3ns but the score was much the same.

    [​IMG]
    [​IMG]
    [​IMG]
     
    Last edited: Oct 27, 2020
  13. shane41

    shane41 Member

    Joined:
    Nov 10, 2008
    Messages:
    6,824
    Location:
    dice que no sabe
    You are getting good at this :)
    ____________

    This is for Mr Chilled on 10 core. 1636. 159 - 143 - 150 - 159 ms in tests

    Default bios. Manual rams. Enable remove limits. 50x core " I set 1.25Vcore " you may want more
    Mem speed 3200. 1.5v you may want less.
    HT off All cores 10. Advanced > disable the x2 speed step things, disable c states. LLC4. That's not much, for big improvement in score.
     
  14. shane41

    shane41 Member

    Joined:
    Nov 10, 2008
    Messages:
    6,824
    Location:
    dice que no sabe
  15. dasa2

    dasa2 Member

    Joined:
    Feb 5, 2003
    Messages:
    841
    Location:
    Northern VIC
    Should be ~8c in the morning so I will give 5.3GHz a crack and see if 2-3 cores can go much higher than 4.

    interesting that at the same clock speed we score the same in the first two tests but you pull ahead in the last two?
     
  16. spit051261

    spit051261 Member

    Joined:
    Sep 29, 2019
    Messages:
    2,810
    Location:
    Perth
    Cpu speed seems more important than ram .
    Interesting.
    Ill try balls for leather and 57 to 58 @ 4800 CL 14 1T
    On an i9 9900k :rolleyes:
     
    Last edited: Oct 27, 2020
  17. dasa2

    dasa2 Member

    Joined:
    Feb 5, 2003
    Messages:
    841
    Location:
    Northern VIC
    Log and Exp time seems high vs 10700k on HWbot at 5275MHz. CPU difference or a tweak?

    [​IMG]
     
    Last edited: Oct 28, 2020
  18. shane41

    shane41 Member

    Joined:
    Nov 10, 2008
    Messages:
    6,824
    Location:
    dice que no sabe
    This is what I wanted to show you. I deliberately broke the third test. 1st try with a hard profile of mine


    Click to view full size!
     
  19. spit051261

    spit051261 Member

    Joined:
    Sep 29, 2019
    Messages:
    2,810
    Location:
    Perth
    1899
    Here ya go Ginger, best I can do :leet:
    Intel 10900k @ 5.9 , Apex XII and ram at shitty timings and speed.
    I will try some other CPU's for you ( 8 and 6 core maybe).
    [​IMG]
     
    Last edited: Nov 1, 2020
    ginger_nuts likes this.
  20. shane41

    shane41 Member

    Joined:
    Nov 10, 2008
    Messages:
    6,824
    Location:
    dice que no sabe
    World record darth maul :D
     
    spit051261 likes this.

Share This Page

Advertisement: