Results 1 to 5 of 5
  1. #1
    New ROGer Array
    Join Date
    Mar 2020
    Reputation
    10
    Posts
    3

    Trying overclocking Asus ROG Strix Z390-H+i7-9700K

    I've build the following system:

    MB: Asus ROG Strix Z390-H
    CPU: i7-9700K (8 Cores up to 4.9 GHz Turbo unlocked LGA1151 300 Series 95W)
    RAM: HX432C16FB3K2/32 HyperX Fury 32GB 3200MHz DDR4 CL16 DIMM (Kit of 2)
    Water cooler: CORSAIR iCUE H100i
    Storage: 2 Crucial P1 500GB 3D NAND NVMe PCIe M.2 SSD - CT500P1SSD8 (RAID0)
    Display card: None (probably some 1070i later)

    I'm also using some of my old computer parts:
    Box Antec P280
    Power supply Antec 380 (going to replace it with something with more power ASAP)

    I've started with the BIOS automatic over-clocking option, and got some overwhelming results: 5.2 GHz with CPU Core Voltage of 1.429. Not surprisingly, running Intel burn test caused system restart so I quickly returned back to the 'safe default'

    Currently, I'm running my system with the following parameters:
    Multiplayer: 48
    VccCore: 1.275 manual
    LLC: 6

    Running Intel burn test in high mode passed successfully several times and uses about 110W according to CoreTemp (and about 180W measured from the mains) with temperature about 75 Celsius. I'd like to improve the over all performance (speed and/or power) and I hope someone will know how to continue

    1. What is the highest temperature I might reach continually without damaging the CPU? The system is 24/7 (not fully loaded of course) and I intend to keep it for the next 3-4 years.
    2. The same question about voltage, what is the highest voltage recommended for using without shortening (too much) the system working life time?
    3. I got temperature spikes when running Intel burn test, a different core reach 85+ Celsius for a short (about a second) time. Why that thing happening? Does it might cause trouble/problem?
    4. I'm using HWMonitor 1.41 and Core Temp 1.15.1 for reading the system status, the temperature reading are differ in the high range and become 10 more in HWMonitor, what software is the accurate one?
    5. I using VccCore set to 1.275 in manual mode and its the CPU voltage when it unloaded. but as soon as pass 25% the voltage starting to rise up. what the reason for it?
    6. Is there a way to connect the iCUE H100i to be controlled by the bios? currently, because the pump controlled using USB, I can't control any temperature profile unless I use the iCUE software (which a plain peace of crap, trust me).
    7. Is the ASUS windows based over-clocking SW really work? I've tried it and it seems like most of the parameters can't be change...
    8. Is there a quick over-clocking guide for the ROG Strix Z390? I don't know what to change accept from the basic.
    9. Is there a way to do both over-clocking and still use the automatic frequency change the CPU normally use for power saving? After all, most of the time I don't use the CPU more than 20%
    10. Is there a way to switch between bios setting without restarting the system? for now I hibernate windows, switch profile in the bios and let the system to continue back out of hibernate. It doesn't very useful mode...

    Thanks in advance, Haim

  2. #2
    ROG Guru: Yellow Belt Array Carlyle2020 PC Specs
    Carlyle2020 PC Specs
    MotherboardAsus Maximus XI APEX
    Processorgaming9900ks@5.2/avx 0/ adaptive vcore / 1.35+0.020 offs/LLC5/ac ll 0.01/TVB off/power saving on
    Memory (part number)Gskill 4400 CL16 /65k r /65k w / 60k c
    Graphics Card #13080 ftw ultra /pads on backplate
    Sound CardXonar STXII mods:3xLME 49720NA
    MonitorLG 38GL950G / -HDR / +madVR
    Storage #12TB 970evo plus nvme
    Storage #21TB 960evo nvme +2x 1tb evo 860 hw-sata-raid
    CPU Coolerarctic freezer II 280
    CaseUnique Silverstone FT2 dampened up the wazoo V2.0
    Power SupplySeasonic TX-850
    Keyboard Cherry KC 1000 on top of a cherry MX1 for ergonomics :)
    Mouse MX Master 3
    Headset Sennheiser PC360 + Philips X2HR
    Headset/Speakers A250 / cause LG was stingy
    OS Win 10 64bit
    Network Routeryeah right...
    Accessory #1 Silverstone EBA01 headphone stand / black
    Accessory #2 radio Keyfob to turn on and off
    Accessory #3 4 bay fantec /10tb red/+scrap

    Join Date
    Mar 2018
    Reputation
    17
    Posts
    155

    Hey there Haim,

    welcome. That is a lot to ask for people to answer 10 questions. But i will try:

    1. What is the highest temperature I might reach continually without damaging the CPU? The system is 24/7 (not fully loaded of course) and I intend to keep it for the next 3-4 years.

    Only 3-4 years? That's short in my book.
    You can easily clock it like you stole it for that time without worry.
    = 95C or 203F in your described personal usecase is my suggestion. 5% headroom already included.

    5.2 will be the stable roof i personally believe

    2. The same question about voltage, what is the highest voltage recommended for using without shortening (too much) the system working life time?

    Keep it under 1.4 volt long term.

    3. I got temperature spikes when running Intel burn test, a different core reach 85+ Celsius for a short (about a second) time. Why that thing happening? Does it might cause trouble/problem?

    No. Use LLC5 and slightly higher voltage to slightly lower them by having less overshoot.
    ...maybe watch a tutorial on how to apply thermal paste to be on the safe side.

    4. I'm using HWMonitor 1.41 and Core Temp 1.15.1 for reading the system status, the temperature reading are differ in the high range and become 10 more in HWMonitor, what software is the accurate one?

    In general you already know and use the rule: "Always pick the higher readout to be safe"
    I personally trust the HWinfo64 readouts the most.

    5. I using VccCore set to 1.275 in manual mode and its the CPU voltage when it unloaded. but as soon as pass 25% the voltage starting to rise up. what the reason for it?

    Referring to my suggestion in answer 3. : lower LLC to 5 to minimise voltage overshoot.
    ;-) voltage and heat go hand in hand

    6. Is there a way to connect the iCUE H100i to be controlled by the bios?

    Currently not that i know of.
    ;-) That could be for the suggestion box

    7. Is the ASUS windows based over-clocking SW really work? I've tried it and it seems like most of the parameters can't be change...

    Some parameters are for other chips maybe.
    In my experience the Intel Xtreme Utility software has less options.

    8. Is there a quick over-clocking guide for the ROG Strix Z390? I don't know what to change accept from the basic.

    That one maybe ask separately to attract ROG Strix owners to share . Or use the forum search function which offers a wealth of information.

    9. Is there a way to do both over-clocking and still use the automatic frequency change the CPU normally use for power saving? After all, most of the time I don't use the CPU more than 20%

    Jupp. Should be on in Bios since you use auto i believe. In windows set power saving options to balanced and in the config itself minimal CPU usage to 5%.
    HWmontior should show the cores running between 800-5200 in your case.

    10. Is there a way to switch between bios setting without restarting the system? for now I hibernate windows, switch profile in the bios and let the system to continue back out of hibernate. It doesn't very useful mode...

    Anything that you can't change with the softwares mentioned:
    hibernate or restart.

    There are a lot of good overclocking manuals out there !

    Best Regards
    Carlyle
    Last edited by Carlyle2020; 04-30-2020 at 11:11 PM.

  3. #3
    New ROGer Array
    Join Date
    Mar 2020
    Reputation
    10
    Posts
    3

    Thanks Carlyle

    Something happened to my (hopefully) my power supply and the computer doesn't power up any more I hope to fined myself another power supply this weekend and see if the system will work.
    Thank you for the voltage/temperature limits, that was my main concern, working in 85 will let me pass the 5GHz without to much work...

  4. #4
    New ROGer Array
    Join Date
    Mar 2020
    Reputation
    10
    Posts
    3

    Some progress...

    Being stuck at home, due to the current situation(...), I kept trying to fix the problems.
    The power supply was changed to PowerBird 1300W and I disassembled the computer and now its laying in the table with only one card assembled (RAID controller), I'm using the board graphic card with two monitors and it work pretty good. I keep having the stopping problem during boot, the green LED keep lighting (meaning its in boot state and that's it) and nothing else happen. After many On/Off the systems start working again. any one ever had this problem?

    Currently my system set to multiplayer = 49, VCore = automatic and LLC=3. I can run the IntelBurnTestV2 in reaching acceptable temperature (70)
    I'm using CPUID HWMonitor for reading the CPU measurements .

    I'm not sure all the testing/loading application are working correctly, I'm running the system with AVX ratio of 2 (I'm going to use this computer for also video encoding). When running HandBreak converting video (H265 12bit) the clock was stable in 4.7GHz, VCore about 1.4V, the temperature around 70 and the system power was 280W. The frequency dropping didn't appeared.
    Using IntelBurnTestV2 show non-continues measurement on the CPUID HWMonitor, at first the clock is in its highest setting with temperature about 70 degrees and voltage nearly 1.4V but after a few seconds its start declining.
    The clock drop to 3.8GHz and the temperature drop to 50. Is it thermal throttling? How to stop it? after all I don't mind 70 degrees on the CPU.
    Running the FurMark CPU burner seems it stress the system to the maximum. The clock was stable in 4.9GHz, VCore about 1.4V, the temperature was around 80 and the system power was 320W. The frequency dropping appeared after a few minutes but it seems affecting different cores, some cores just popped down momentarily to a lower frequency but than returned back to normal speed and other cores lower theirs frequency.

    -Is there a way to control VCore better? currently, on automatic every thing work fine but VCore reach 1.5V under heavy load and cause a lot of heat.
    -What to do next? I'm running out of ideas how to stabilize the system with lower voltage.
    -Is the testing application are really important in the real world? I try running DIVX Converter (without HW acceleration) and it reach 100% CPU for more than an hour and the temperature was less than 60 at 4.9GHz, with the same settings the testing application fails. This real world application is what I'm interested, I'll be happy if I could trust the system run flawlessly all the time.
    -What is the name of the parameter controlling the clock multiplier according to the load? I used to work with 1.2GHz upon low load and pop up to 4.5 in full load, I cant remember where it was set...

  5. #5
    ROG Guru: Yellow Belt Array Carlyle2020 PC Specs
    Carlyle2020 PC Specs
    MotherboardAsus Maximus XI APEX
    Processorgaming9900ks@5.2/avx 0/ adaptive vcore / 1.35+0.020 offs/LLC5/ac ll 0.01/TVB off/power saving on
    Memory (part number)Gskill 4400 CL16 /65k r /65k w / 60k c
    Graphics Card #13080 ftw ultra /pads on backplate
    Sound CardXonar STXII mods:3xLME 49720NA
    MonitorLG 38GL950G / -HDR / +madVR
    Storage #12TB 970evo plus nvme
    Storage #21TB 960evo nvme +2x 1tb evo 860 hw-sata-raid
    CPU Coolerarctic freezer II 280
    CaseUnique Silverstone FT2 dampened up the wazoo V2.0
    Power SupplySeasonic TX-850
    Keyboard Cherry KC 1000 on top of a cherry MX1 for ergonomics :)
    Mouse MX Master 3
    Headset Sennheiser PC360 + Philips X2HR
    Headset/Speakers A250 / cause LG was stingy
    OS Win 10 64bit
    Network Routeryeah right...
    Accessory #1 Silverstone EBA01 headphone stand / black
    Accessory #2 radio Keyfob to turn on and off
    Accessory #3 4 bay fantec /10tb red/+scrap

    Join Date
    Mar 2018
    Reputation
    17
    Posts
    155

    I would suggest playing around with adaptive voltage for a week.

    Then use the adaptive vcore settings for another week of manual vcore testing.
    That inlcudes trying LLC settings 4-6 with both to see what is the coolest option.
    The testing application in your case would be the divx converter. Ditch Furmark and the Intel Burn Test (which is the more reasonable of the two, but far from good) as well.
    100% load means not much. The power draw still can be heavy or light.
    Check CPU power in HWinfo to see what the divx converter is eating.
    Then check the power draw of Furmark and compare the two to draw your own conlcusions.

    Maybe you are looking for :
    EXTREME TWEAKER -> INTERNAL CPU POWER MANAGEMENT:
    Long Duration Package Power Limit..................4095
    Package Power Time Windows.........................127
    Short Duration Package Power Limit.................4095

    Also :
    CPU Core/Cache Current Limit Max to 255.50

    These settings are max and not really suggested in your case since you have no cooling headroom left.
    They rather help getting higher clocks with lower loads and using lower values to throttle the rig if it's under heavy load.

    So check the interweb for the standard short/long term value for your 9700k.
    Put those in manuall any play around withthem.

    If you have a feel for them use a 5Ghz clock and in your case an avx offset of - 1 to -3.
    Get it stable and then throttle it with the limiters if it is under heavy long term loads.

    Use cinebench R15 for quick tests of new settings
    Use cinebench R20 if you think you have found a setting worth keeping.
    Run it with HWinfo ot see cpu power draw and heat generation.
    It can also show graphs if you double click on a value but you can simply run MSI Afterburner for that since it those graphs in its standard config already. With the graphs you check if the CPU holds 4.9/5.0 Ghz during the whole test.

    The last sanity check would be using the divx converter to see if you are better off than before in (either/or):

    a) heat generation
    b) time to finsih a project

    ps:
    getting the best possible results in cinebench using adaptive vcore usually means running 2-3 runs back to back

    ... "My PSU did not like that at all" - he said, whilst looking at his dead 1000w titanium PSU of last generation.

    So learn from my mistake and give it some seconds inbetween tests if you are going to run cinebench an unreasonable amount of times every day for the foreseeable future
    Last edited by Carlyle2020; 05-01-2020 at 12:13 AM.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •