Clevo 2020

Discussion in 'Sager and Clevo' started by Dakka3, Aug 28, 2019.

Tags:
  1. jc_denton

    jc_denton V̖̟en̰g̻̼̰̩͙ea̲n̪c̭e̼ ͍̘̤͓̟̤Is̙͔ ̤Mi̻̭̣͎͍ͅn̰ḙ

    Reputations:
    5,105
    Messages:
    2,088
    Likes Received:
    3,152
    Trophy Points:
    281
    What about when you compare 210 to 160? On my end the reported VID was higher at 160 (1.305v) compared to 210 (1.245v), yet lower temp and slightly lower package draw.

    @Prema on setting AC/DC to "1"
    http://forum.notebookreview.com/threads/clevo-overclockers-lounge.788975/page-1611#post-10748293
     
    Last edited: Mar 5, 2020
  2. ole!!!

    ole!!! Notebook Prophet

    Reputations:
    2,439
    Messages:
    5,809
    Likes Received:
    3,747
    Trophy Points:
    431
    210 to 160 has little to no change. going down 50 would mean you need to up voltage at least 50mv so voltage reading will always be higher. 8700k was fine at ac/dc value of 1, I couldnt even get it past booting into windows 3/5 times at ac/dc 40 for the 9900ks and would constantly crash, at 60 it was more stable.

    210 - 60 would mean i'd need to up the voltage around 150mv but resulting in lower temp by only ~3C. im sure with using 2nd fan + max fan this would come to ~5C or more. temp difference will only show when its really stressed.

    I did find out something interesting maybe @unclewebb and @Mr. Fox could explain. changing the voltage value in TS under cache instead of core has no effect on load voltage instead has effect on idle voltage/c-state. normally voltage would be around 0.75v when on idle, now its 0.6v when its clocked down. this is great for battery use only using ~6w for 8 cores lmao.
     
  3. Mr. Fox

    Mr. Fox BGA Filth-Hating Elitist

    Reputations:
    31,673
    Messages:
    36,092
    Likes Received:
    58,831
    Trophy Points:
    931
    Yes, I have the same observation. But, I do not have an explanation for it. I can also set a much higher negative offset on core voltage, as long as the cache voltage is not set too low. It's like the cache voltage controls the bottom end of the voltage rather than the top end.
     
  4. jc_denton

    jc_denton V̖̟en̰g̻̼̰̩͙ea̲n̪c̭e̼ ͍̘̤͓̟̤Is̙͔ ̤Mi̻̭̣͎͍ͅn̰ḙ

    Reputations:
    5,105
    Messages:
    2,088
    Likes Received:
    3,152
    Trophy Points:
    281
    I think reason I saw -5c lower max temps was that at '210' offset was -120mV, while at '160' offset was -100mV. Shame we don't have vcore to see what's going on behind the curtains.
     
    Spartan, Mr. Fox, ole!!! and 2 others like this.
  5. Meaker@Sager

    Meaker@Sager Company Representative

    Reputations:
    8,658
    Messages:
    54,017
    Likes Received:
    15,654
    Trophy Points:
    931
    The uncore which I believe it also covers does behave differently to the core in terms of clocking. There could also be voltage difference limits.
     
    Mr. Fox likes this.
  6. jc_denton

    jc_denton V̖̟en̰g̻̼̰̩͙ea̲n̪c̭e̼ ͍̘̤͓̟̤Is̙͔ ̤Mi̻̭̣͎͍ͅn̰ḙ

    Reputations:
    5,105
    Messages:
    2,088
    Likes Received:
    3,152
    Trophy Points:
    281
    vcore/cache share the same rail on the 9900kx from my understanding.

    With override voltage you could do "1", that way the VID would be equal to set voltage, I tried that with the 8086k. But the 9900K does not play well with override voltage and will trigger hard shutdown under load. Probably hitting the hw amperage limit.

    Edit:

    Some more testing with Fire Strike Physics 15min loop 5Ghz x47

    AC/DC 210
    Offset: -130mV
    Max temp: 93c
    VR VCC: 55c
    Package power draw: 130.461W

    AC/DC 160
    Offset: -100mV
    Max temp: 88c
    VR VCC: 54c
    Package power draw: 126.671W

    Only change at 210 LLC was I was able to increase the undervolt to -130mV from -120mV and still keep stability in other applications.
     
    Last edited: Mar 6, 2020
    Spartan and Mr. Fox like this.
  7. Falkentyne

    Falkentyne Notebook Prophet

    Reputations:
    7,765
    Messages:
    5,674
    Likes Received:
    7,984
    Trophy Points:
    681
    Override voltage on laptops controls "VID", not "Vcore", but laptops use adaptive voltage regardless of what the option says. The "Adaptive" voltage setting uses Intel vCPU (base VID), while the "Override" setting reprograms base VID to the override setting.

    vCPU is VID when ACLL and DCLL are both set to 0.01 mOhms (1)
    LLC=Loadline Calibration from VRM (usually 1.6 mOhms on 8 core and 2.1 mOhms on 6 core)
    Set DCLL=ACLL=0.01 and remember Vcpu (Vcpu=VID in this case) in idle and low/high load
    Set DCLL=0.01 ACLL=LLC and remember ACLL droops (ACLL_droop=VID-Vcpu from (1)) in idle and low/high load
    Now we can calculate final VID:
    VID=Vcpu-DCLL*I+ACLL_droop

    VID=Vcpu+ACLL*dI (dI=I1-I0)

    If ACLL=LLC (Loadline calibration):

    VID=Vcpu-DCLL*I+ACLL_droop
    Vcore=VID-vdroop+offset

    Vcore=Vcpu-DCLL*I+ACLL_droop-LLC*I+offset.

    Offset mode:
    VID=Vcpu-DCLL*I+ACLL*dI
    Vcore=VID-LLC*I+offset

    It seems, on these laptops, while VID uses this formula, final vcore ignores the DC Loadline value.
    Seems to be the same way on Gigabyte Z390. VR VOUT ignores DC Loadline value.

    I think only Asus boards use both AC and DC Loadline to determine vcore in offset mode while most other systems only use AC Loadline and LLC and voffset.
     
  8. jc_denton

    jc_denton V̖̟en̰g̻̼̰̩͙ea̲n̪c̭e̼ ͍̘̤͓̟̤Is̙͔ ̤Mi̻̭̣͎͍ͅn̰ḙ

    Reputations:
    5,105
    Messages:
    2,088
    Likes Received:
    3,152
    Trophy Points:
    281
    Since mobile 20x0super series is launching right around the corner, I think it's safe to assume that the x170 is shipping with this option.
     
    Eclipse251 and Spartan like this.
  9. gooface

    gooface Notebook Evangelist

    Reputations:
    222
    Messages:
    518
    Likes Received:
    126
    Trophy Points:
    56
    I really hope sager comes out with something comparable to the Asus ROG Zephyrus G14 with the 4900HS or the 4800.

    The reviews make anything intel has out right now look like trash. (unless you get an extremely thick and expensive laptop.
     
  10. yrekabakery

    yrekabakery Notebook Virtuoso

    Reputations:
    924
    Messages:
    2,738
    Likes Received:
    2,581
    Trophy Points:
    181
    I really hope not. The G14 is undesirable to me:
    • One RAM module is soldered, so you can't upgrade the terrible 3200MHz CL22 it comes with
    • 120Hz IPS panel is disgustingly ghosty at 22ms GTG response time
    • CPU-limited gaming performance trails the i7-9750H
    • Has only 1/4 the L3 cache as desktop Zen2
    • GPU is Max-Q
    • CPU/GPU temps aren't great despite the low TDPs
     
    Last edited: Mar 30, 2020
    ha1o2surfer, raz8020 and Papusan like this.
Loading...

Share This Page