ClockTuner for Ryzen (CTR) - and introduction guide and download

Discussion in 'Processors and motherboards AMD' started by Hilbert Hagedoorn, Sep 29, 2020.

  1. haps

    haps New Member

    Messages:
    1
    Likes Received:
    0
    GPU:
    R7870@1300
    Hi,
    I've been using ClockTuner for like 3 weeks. It set my Ryzen 3600@1325 mV at 4450/4425 (CCX0/1) - cooled by custom WC, old cpu block I have since Barton 2500. I've been gaming alot, no single crash even with cpu demanding games like Kingdome Come played for hours. Cinbench temp is up to 70C. Gaming up to 50C
    Today I tried Prime95 small FFT and every time it instantly sprays errors, I think its because temp also instantly reaches 88C, that is max set in Clocktuner. Is is "normal" behavior ?
    Temp does not build up. Cinebench test is similar, it starts like 67C and goes max to 70 on longer run.
     
  2. Efreeti

    Efreeti Member

    Messages:
    10
    Likes Received:
    0
    GPU:
    EVGA 3080 FTW3Ultra
    Reference voltage was 1250 mV and reference frequency was 4275 MHz
    Max frequency says 4675 MHz and diagnostic voltage at 1231 mV, max PPT 200 W, max EDC 190 A.

    Going roughly by your advice I tried running it with reference voltage set to 1275 mV and reference frequency down to 4200 MHz, with Enhanced Accuracy also enabled.
    It got much much farther than before, but still crashed my system at Step 25. From Step 20 and onwards, the progress bar for each stress test just stopped progressing at about 35% every time. The stress tests still ended, but it never reached the end of the progress bar.
     
  3. sakaana

    sakaana New Member

    Messages:
    4
    Likes Received:
    3
    GPU:
    5700XT
    Beta 4 should be available soon today - source: Twitter :)
     
    Efreeti and M4xweLL like this.
  4. mannix

    mannix Active Member

    Messages:
    70
    Likes Received:
    18
    GPU:
    MSI GTX1070 ArmorOC
    Hey @1usmus I did sent you a PM, hope you'll read it :)

    There's a small bug in the logic how the voltage and frequency is applied.
    That's why during the diagnostic or manual applying a profile you get a sudden reboot.

    In layman terms, moving from setting at 4.4GHz and 1.3v to a 2.2GHz and 1.0v you have first to set the clock and then the voltage.
    CTR right now is always setting first voltage and then clocks; meaning for a very short time the CPU will be set at 4.4GHz and 1.0v.
    Which is a safe bet for reboot.

    Also during diagnostic the frequency is fixed and voltage scaled down.
    But at some point the bad binning like mine they just crash.
    Hopefully he'll find a better way; like scaling up or scaling down a bit the clock before going down with voltage and test if it's unstable.

    It's a wonderful software and it will get even better!
     
    1usmus likes this.

  5. ST4H

    ST4H Member

    Messages:
    10
    Likes Received:
    0
    GPU:
    5700XT 8GB
    Just checked on versions 1.0 Beta 3 and 1.1 Beta 4. It's the same everywhere so its not a recent bug
     
  6. mibo_ger

    mibo_ger New Member

    Messages:
    7
    Likes Received:
    1
    GPU:
    rx480
    https://mobile.twitter.com/1usmus/status/1317873168695046145?p=v

    1usmus seems to be aware of the 3300X issue.
    I'm a bit surprised that a few 3300X show up in the google docs CTR list. So, some 3300X do actually work with CTR?
     
  7. 1usmus

    1usmus AMD Memory Guru

    Messages:
    50
    Likes Received:
    98
    GPU:
    Nvidia GTX 1080Ti
    Hi Mate!
    I read your message, thank you. At the moment there is a registry poll asking if a command was executed otherwise the next command will not be sent.
    But in any case I will check the information you have provided.

    BR, Yuri
     
  8. 1usmus

    1usmus AMD Memory Guru

    Messages:
    50
    Likes Received:
    98
    GPU:
    Nvidia GTX 1080Ti

    At the moment I don't know why some users have a problem with 3300X. It's either old drivers or an incompatible BIOS.
     
  9. Efreeti

    Efreeti Member

    Messages:
    10
    Likes Received:
    0
    GPU:
    EVGA 3080 FTW3Ultra
    As someone who still has never had CTR finish without the reboot/crash, I hope that something can still be done.
     
  10. Remekra

    Remekra Active Member

    Messages:
    85
    Likes Received:
    6
    GPU:
    AMD Radeon RX6900XT
    3900X and ASUS X570 TUF Gaming here with 240 AIO Cooler.
    I have ran CTR 1.1 for the first time. According to diagnostics I have a golden sample. It started to overclock it, seemed stable but after Cinebench has started I've had a restart. Ran again but this time with LLC set to Level 3 and Delta set to 150MHz. It ran fine, Cinebench has passed, score I received was a little above 7700 (stock was just shy of 7100). Run stability test, also no problems.
    Frequencies are as follows:
    CCX1 and CCX2: 4400MHz. CCX3: 4375MHz, CCX4: 4325MHz.
    VID is set to 1250mV.

    First I thought it was great. Before with PBO and AutoOC in BIOS I've had around 7400 points in Cinebench, and temps were running around 80. Now the temps were hovering around 60-70C and I have more performance. But after a restart I've run Cinebench once more and it started to either crash or restart the PC completely.
    Changed LLC to Level 4, ran Cinebench many times, crashes have stopped. Then I thought I will run Blender as I have it. First Junk Shop scene. It ran great, temps a little over 60 (was 80 when in PBO). Then I've loaded Spring scene (can be downloaded from blender website) and crashes have returned.
    I noticed that when Blender (or Cinebench when in LLC3 mode) is rendering a tile that is harder to render it runs fine, but problems start to happen when in a scene like Spring there is a lot of background that can be rendered quickly temps jump to around 73C and it crashes on those parts of render. Which is strange as they are lighter and CPU should be doing them fine.
    I've tried LLC5 but the same happens. I changed it back to LLC4, Cinebench still running fine, also tried Corona Benchmark also fine. Ran 3D Mark Time Spy Extreme, also no crashes temps around 60C. Ran Mafia Definitive edition and RDR2 also no crashes so far (and temps are amazing from 40 to 50, before with PBO it could hit 70 sometimes).
    Seems to me that problem happens when CPU is loaded heavy (like rendering tiles in Blender that are hard to do) and then it swifts over to lighter task (background tiles in Blender) temps rise (but never above 75C) and then it crashes. It also could be seen in Cinebench when in LLC3, as some parts of that render are also more heavy to do and some are lighter.
    For now I will leave the settings that I have to see if any crashes will happen when gaming or otherwise, but any advise on what could be happening?
     

  11. xNako

    xNako New Member

    Messages:
    3
    Likes Received:
    0
    GPU:
    RTX 2060
    Hi guys,

    Why have i too high vdroop?
    https://imgur.com/a/kDIejTj

    R5 3600
    MSI B450 Mortar Max (LLC 3)
    Arctic Freezer 34 eSports Duo


    Regards.
     
    Last edited: Oct 20, 2020
  12. ST4H

    ST4H Member

    Messages:
    10
    Likes Received:
    0
    GPU:
    5700XT 8GB
    Bios version 7C35vA6 (latest stable) on msi x570 unify is the same as newest beta version, so i don't think it's bios problem
     
  13. Garwinski

    Garwinski Member Guru

    Messages:
    185
    Likes Received:
    4
    GPU:
    XFX Fury X
    Maybe up LLC a bit? Or have you tried it wit LLC set to auto? CTR works with auto now as well.



    Btw, great program! Before, I only saw maybe one core peak at around 4250mhz with PBO enabled. Now, I have all cores at 4425mhz 1250mV (Ryzen 3600, Gold Sample).

    I can also take these settings and put them in my BIOS, right? For PPT, EDC, and TDC I will have to set PBO to manual and just input the numbers. But what do I do for Voltage Scaler? When I enable PBO, this gets enabled as well, but I cant disable it. I can set it on auto, which I dont want, or from 1x to 10x. Is 1x the same as disabled?
     
  14. Thelgow

    Thelgow Member

    Messages:
    15
    Likes Received:
    2
    GPU:
    GTX 1070
    Thanks 1usmus. Some minor issues. I started on 1.1b3 and diagnostic ran fine, said golden. Then ran the test, and for my 3600 it said 4475/4450 1.256v. My only issue was it wasnt applying profiles at start, or even saving the profile properly. apply would be grayed out so I had to do save then apply.
    Trying 1.1b4 and things a little different. I ran diagnostic which was fine, however upon clicking copy to clipboard it crashed. I relaunched it but have since realized it keeps the log in the folder but seems to clear it on launch. No biggy. I run the tuning again and on test3 it seems to have crashed my UPS. this is odd as Im on a 600w ups and all during the test it stayed under 280. Even when gaming with my gtx1070 and it ramps up it only hits 480.
    So I ran the tuning again and this time its fine. It did settle on 4475/4425 which is 50mhz less on ccx2, but no biggy.
    Side note Ive noticed cinebench likes to run in below normal priority. I set to high and scores jumped a bit too.
    Pre CTR - CB20 score 3433. Although I have screenshots of cb20 ranging from 3064 to 3617 when i was playing with dram OC's. I have it stock currently.
    1.1b3 bumped it to 3811.
    1.1b4 started with 3571 got it up to 4029. Rerunning cb20 manually 1 more time with High priority yielded 4047. Good work. So from a fluctuating 3060-3430 to 4047.
    Ill let this sit for a bit, then back to Memory OC'ing.
     
  15. Wanya

    Wanya New Member

    Messages:
    9
    Likes Received:
    2
    GPU:
    ATI
    how did single core performance change for you ?
     

  16. Kinetic_BLR

    Kinetic_BLR New Member

    Messages:
    2
    Likes Received:
    0
    GPU:
    1660 ti
    Hi,

    I have a question, is the issue with profiles auto load still exists in beta 4? I've tried and there is no problem with CTR autostart, but profiles are not applied on Win start.

    And one more question, if I previously used beta 3 and I know the value for VID, CCXs can I just write them into beta 4 Profile, save and use? Or there is some other "behind scene" whuch is recorded to some configs and I need to redo Diag -> Start?

    Thanks!
     
  17. Kaltern

    Kaltern Active Member

    Messages:
    53
    Likes Received:
    5
    GPU:
    Gigabyte 2080 Super
    Date & Time User name CPU name Motherboard Cinebench R20 VID voltage
    20.10.2020 19:26:50 @kalte AMD Ryzen 7 3700X 8-Core Processor ASUS TUF GAMING X570-PLUS 3797 1.256
    PPT CCX1 CCX2 CTR ver. Diag. voltage Diag. frequency
    99.6 4175 4200 CTR 1.1 beta 4 1130 4050

    Something is very wrong here... Suggestions would be most welcome. My previous use of CTR 1.0 was far more favourable, and I don't believe anything has changed....
     
  18. Roccer

    Roccer Ancient Guru

    Messages:
    1,944
    Likes Received:
    7
    GPU:
    EVGA 2080Ti XC ULTR
    1.1b4 resulted in hard reboot during the main run (diagnostic run was fine). somewhere around step 10 or so (i had walked away and didn't see it). event viewer reported this error:

    A fatal hardware error has occurred.
    Reported by component: Processor Core
    Error Source: Machine Check Exception
    Error Type: Cache Hierarchy Error
    Processor APIC ID: 5
    The details view of this entry contains further information.
    For now I went back to 1.1b3 and it's profile as that suits my needed. may play with 1.1b4 (and up voltage a bit) and try again, but that's for another day.


     
  19. M4xweLL

    M4xweLL New Member

    Messages:
    4
    Likes Received:
    1
    GPU:
    MSI GTX 1070 GAMING
    I have the same issue, autostart CTR it's OK but profile aren't applied on Win Start!. Any idea please?

    Thanks!
     
  20. WhiskeyOmega

    WhiskeyOmega Master Guru

    Messages:
    273
    Likes Received:
    22
    GPU:
    RTX 3090 MSI GXT
    Ive given up using this.
    Its far too unstable could potentially do damage from instant reboots and doesnt offer accurate stress testing between the level difference from prime95 to CB20
     

Share This Page