The AMD Ryzen All In One Tread /Overclocking/Memory Speeds & Timings/Tweaking/Cooling Part 2

Discussion in 'Processors and motherboards AMD' started by vbetts, Sep 24, 2018.

  1. OnnA

    OnnA Ancient Guru

    Messages:
    13,511
    Likes Received:
    3,441
    GPU:
    3080Ti VISION OC
    Lowered latency :p for FREE.
    Setting power plan setting interrupts to "proccesor 1" it can decrease memory latency by around 0.8ns

    Download this tool (im using it always to set my own Power Plan configuration, 1Usmus Plan with my Tweaks :cool:)
    (Original Post -> https://forums.guru3d.com/threads/windows-power-plan-settings-explorer-utility.416058/ )
    -> http://www.mediafire.com/file/wt37sbsejk7iepm/PowerSettingsExplorer.zip/file
    Please unhide: Interrupt steering.

    Note:
    Tested & it's working, but everyone need to check the perfect option for themself ;)
    I have it set at - <Lock Interrupt Routing> now to proccesor 1 (it gives me ~0.3ns)
    4x8GB is used.

    Source: https://www.overclock.net/forum/showthread2.php?p=28201630#post28201630

    [​IMG]
     
    Last edited: Nov 20, 2019
  2. CyberSparky

    CyberSparky Member Guru

    Messages:
    123
    Likes Received:
    75
    GPU:
    EVGA RTX 3080 XC3
    Thank you for this, tested and stable to 800% MemtestPro, and 2.5 hrs of BFV.

    I am currently running B-Die, ProcODT 36.9- 24-20-20-24, disabled/disabled/RQZ-5/ 16-17-16-16-32-48-288 @ 1.47v getting 63.0-62.8 latency with SMT enabled and NO SMT 63.9-63.8, I stole/added your 1-4-4-1-2-2-1 and it dropped my latency from 63.8- 63.3 down to the above.

    Before Agesa 1004, I was never able to go lower than ProcODT 40 or it would never boot, don't know if that had much of an effect vs the timings though. Hats off to your persistence good sir.
     
  3. Here's my logic -
    • Isolate all but one ram module
    • Establish your desired timings
    • Run MemTest86 off bootable DOS environment
    • Using one module; you can tweak, test, pass - faster
    • Verify established stable timings in DOS & Windows
    • Verify end-result simultaneously on all DIMMs
     
    Last edited by a moderator: Nov 19, 2019
    386SX and OnnA like this.
  4. OnnA

    OnnA Ancient Guru

    Messages:
    13,511
    Likes Received:
    3,441
    GPU:
    3080Ti VISION OC


     
    Last edited: Nov 19, 2019

  5. 386SX

    386SX Maha Guru

    Messages:
    1,170
    Likes Received:
    1,241
    GPU:
    AMD Vega64 RedDevil
    That's almost how I already do it, but I use my 2 sticks all the time. I found Ryzen behaves VERY sensitive regarding voltage for example, and what runs with one stick isn't guaranteed to run with two. I know this takes longer for a run to complete, but then I have the end result faster because I can skip the second test.

    In addition the thermal issues didn't show on 1 dimm ...

    So far I think CADBUS / RTT / ProcODT seems to be the issue, as I still lack a reliable info about how to calculate those. IF, THEN, ELSE or the like.

    So far: the higher you set it, the more latency you create. ALL of them do that.

    Does anybody know some "rule of thumb" regarding memtest errors? Like "if error 10 (bit fade) happens, raise voltage of either SOC or RAM." It would make life easier.
     
    Deleted member 213629 likes this.
  6. Clouseau

    Clouseau Ancient Guru

    Messages:
    2,734
    Likes Received:
    428
    GPU:
    ZOTAC AMP RTX 3070
    The power steering setting was cool. Setting it to Processor 1 was the only setting that made a noticeable difference in benchmarks. Games, could not tell a difference. The one thing it did do, made the ipc for single threads look amazing. Need to see if it is noticeable in older games. Single thread CPU-Z results went from 545 to 550; 545 was touch and go while 550 is solid (549 was the lowest result out of ten). Ram latency remained the same at 62.5ns.
     
    OnnA likes this.
  7. Kool64

    Kool64 Maha Guru

    Messages:
    1,115
    Likes Received:
    440
    GPU:
    Gigabyte RTX2070S

    On my 3700X I gained 8 points in CPUZ single to put it at 531 and I did in fact get lower latency by just about .8 on the 1Usmus plan with "procesor 1" Locked didn't have any effect.
     
    OnnA likes this.
  8. OnnA

    OnnA Ancient Guru

    Messages:
    13,511
    Likes Received:
    3,441
    GPU:
    3080Ti VISION OC
  9. Kool64

    Kool64 Maha Guru

    Messages:
    1,115
    Likes Received:
    440
    GPU:
    Gigabyte RTX2070S
    I'm glad I'm not crazy and 1usmus confirmed the lower clocks with the 46.54 SMU.
     
  10. SpajdrEX

    SpajdrEX AMD Vanguard

    Messages:
    2,852
    Likes Received:
    1,131
    GPU:
    Sapphire RX 6800XT
    Is that usable for Ryzen 2000 series?
    Because when enabling it, MemBench fails after few seconds.
     

  11. erouz

    erouz Member

    Messages:
    33
    Likes Received:
    3
    GPU:
    1070ti
  12. Clouseau

    Clouseau Ancient Guru

    Messages:
    2,734
    Likes Received:
    428
    GPU:
    ZOTAC AMP RTX 3070
    erouz likes this.
  13. Kool64

    Kool64 Maha Guru

    Messages:
    1,115
    Likes Received:
    440
    GPU:
    Gigabyte RTX2070S
    I was using The Stils modified BIOS before the ASRock beta. I actually liked it a bit better it had a more aggressive SMU.
     
    erouz likes this.
  14. Webhiker

    Webhiker Master Guru

    Messages:
    737
    Likes Received:
    254
    GPU:
    EVGA GTX 1080Ti SC2
    erouz likes this.
  15. erouz

    erouz Member

    Messages:
    33
    Likes Received:
    3
    GPU:
    1070ti
    Unfortunately there is no modified bios for me from Stils.
    Tks for all advice's
     

  16. Mr. Sunshine

    Mr. Sunshine Master Guru

    Messages:
    236
    Likes Received:
    75
    GPU:
    SapphireNitro V64
    Have all these set as you do. Still only seeing 4.4ghz max boost on 2 cores. In game max I get is 4.350-4.375ghz. Dont know how you guys are getting higher clocks

    I know gerardfraser has that great 3800X combo going which makes sense for higher boost. And you've stated your best boost was 4.575 with steady clocks of 4.425 and 4.450. What CPU are you running? Thought it was 3700X....If its a 3800X then that would explain it.
     
  17. OnnA

    OnnA Ancient Guru

    Messages:
    13,511
    Likes Received:
    3,441
    GPU:
    3080Ti VISION OC
    Maby try different settings?
    CPU 0 or First avaible?
     
  18. Clouseau

    Clouseau Ancient Guru

    Messages:
    2,734
    Likes Received:
    428
    GPU:
    ZOTAC AMP RTX 3070
    Running a 3800X but you are getting the same equivalent boost clocks, actually a bit better.
     
  19. It... performed better for me so far. I'm remoted into it while running some background tasks (performing partition recovery from raid array)[​IMG]
     
  20. gerardfraser

    gerardfraser Ancient Guru

    Messages:
    3,343
    Likes Received:
    763
    GPU:
    R9 290 Crossfire
    Well that is nice to read.If you get a chance ,do you have actual gaming benchmarks and post a comparison,that would be interesting but there should be no difference at all.
     
    Deleted member 213629 likes this.

Share This Page