EVGA SC17 1080 finally available

Discussion in 'Other Manufacturers' started by Vistar Shook, Aug 11, 2017.

Tags:
  1. Vistar Shook

    Vistar Shook Notebook Deity

    Reputations:
    810
    Messages:
    1,000
    Likes Received:
    921
    Trophy Points:
    131
    Not at home. But temps are in the 90s. Don't plan on running at those clocks and voltage again.

    Enviado de meu Pixel 2 usando Tapatalk
     
    Papusan likes this.
  2. Falkentyne

    Falkentyne Notebook Virtuoso

    Reputations:
    3,480
    Messages:
    3,415
    Likes Received:
    4,256
    Trophy Points:
    331
    @Vistar Shook

    So that means you have vdroop then. 1.52v actual voltage rather than VID would reach 100C in <15 seconds.

    Because when I set 1.380v @ 4.9 ghz on mine, I get 90C temps after a few CB runs then the EC shuts the laptop off !

    Found out apparently MSI has a built in "loadline calibration" I found out by changing the "DC Loadline" setting from 1 to auto and I noticed that the "voltage" (not real voltage!!! THERE IS NO VCORE SENSOR IN THESE LAPTOPS!! The VID !!!) would drop at load, the higher the load, the more the drop. I don't know if the drop corresponds to what the vdroop is 'supposed' to be, but it does seem like it, but it's very illogical. it's almost like the setting is bugged.

    Default DC loadline (2.10 mOhms or auto) shows VID droop.but DC loadline at 0.01 mOhms shows NO VID droop at all, and it follows the approximate override voltage set. And if you set IA AC DC loadline to a higher value, the VID gets boosted by the amount of mOhms of resistance (I set 1.275v voltage, AC loadline=2.10 mOhms (Auto), and DC loadline=1, then i started prime95 blend (AVX disabled) at 4.7 ghz and the VID showed as 1.4v with 78W power draw for 2 seconds, temps were high (MUCH higher than the power draw would suggest) and the laptop shut off. (remember when AC loadline is set higher than auto, the power draw seems to be reported too low). But that means 1.275v + AC Loadline=2.10 mOhms = 1.4v true voltage.

    However if AC loadline = 1 and DC loadline=Auto, 1.275v voltage, then the VID is 1.18v at load (!) but the power draw is the same (actually about 4 watts higher) than AC loadline=1 and DC loadline=1 (1.275v showed up as 1.279v VID here) together. So clearly the 'VID droop" isn't real voltage droop.

    Mind if i ask u some questions when ur back at your PC? :)
     
    Last edited: Apr 15, 2018
    Vistar Shook likes this.
  3. Falkentyne

    Falkentyne Notebook Virtuoso

    Reputations:
    3,480
    Messages:
    3,415
    Likes Received:
    4,256
    Trophy Points:
    331
    Edited my post @Vistar Shook and @Papusan to say the VID would drop at load (by almost 90mv) if I set DC Loadline to 0 (Auto, or 2.10 mOhms) rather than 1. But the temps would actually be about 2C *HOTTER* than if DC loadline were set to 1 (0.01 mOhms)! So the reported VID is 'vdrooping' by what a normal desktop system would 'droop with' if Loadline Calibration were disabled on a desktop board. But MSI seems to have a built in LLC (even if AC loadline is set to 0.01 mOhms) with no way to disable or prevent it. I mean, that is a good thing, but it seems like the DC loadline value was designed around having vdroop active. So if I'm guessing right, desktop boards have "2.10 mOhms" of vdroop applied (negatively) to the loadline slope (This is a guesstimation. .i could be totally utterly wrong and stupid), and the IA AC DC loadline setting for 2.10 mOhms "counteracts" this by boosting the VID (which is supposed to be applied to be used for adaptive voltage). But the DC setting seems to be doing the opposite of what you THINK it should do. That's why on MSI boards, if you have the IA AC DC loadline at 2.10 mOhms for both (or auto), the VID that is shown at load is GROSSLY underreported compared to the REAL voltage (because 1) MSI is using built in Loadline Calibration, 2) the IA AC DC loadline setting is BOOSTING the voltage on TOP of the bulit in loadline removing the vdroop to begin with.

    Well I just confused you guys. Have fun.
     
    Vistar Shook likes this.
  4. Papusan

    Papusan BGABOOKS = That sucks!! STAHP! Dont buy FILTH...

    Reputations:
    10,489
    Messages:
    17,635
    Likes Received:
    25,517
    Trophy Points:
    931
    I expect load voltage was more like 1.420. And still, none of you is able to run your 7820Hk@5.2GHz 1.270v same way as @Danishblunt say is the normal for many of this BGA chips. BTW... You kill the Alienwares with your Cinebench scores. I wonder if @Pete Light want to compete in CBR-15 with his Alienware.
    Have you tried all possible ways? Adaptive volt, balanced power profile etc. Even go back to run IA AC DC loadline default (auto)in different power config setups. Tested with c-states enabled etc. In short test out everything. Maybe you can get a higher score if the chips start up with lower temp.
     
    Vistar Shook and Falkentyne like this.
  5. Vistar Shook

    Vistar Shook Notebook Deity

    Reputations:
    810
    Messages:
    1,000
    Likes Received:
    921
    Trophy Points:
    131
    Yeah the vdroop is evident and explains why I run at much higher reported VID compared to other laptops, although the temps say otherwise. I do have a vboost option in the BIOS but it does nothing, I think it is a setting for another motherboad type. AC DC loadline settings doesn't affect the reported VID. That being said, my paste job is crap, having one core running 5C hotter than the others, but this laptop is a pain to redo at the moment, I will wait for the programmer parts to arrive to redo it and I try to get even temps on cores. At 4.9GHz this system is highly unstable, and the CB runs were lucky, most runs stop in the middle, with one core over 95C.
     
  6. Papusan

    Papusan BGABOOKS = That sucks!! STAHP! Dont buy FILTH...

    Reputations:
    10,489
    Messages:
    17,635
    Likes Received:
    25,517
    Trophy Points:
    931
    Better thermal paste may work better. But it’s Amazing that the Unified Heatsink design can’t suck up the heat from only Cpu when you have almost no load from the graphics.
     
  7. Falkentyne

    Falkentyne Notebook Virtuoso

    Reputations:
    3,480
    Messages:
    3,415
    Likes Received:
    4,256
    Trophy Points:
    331
    Running at 5C above other cores if you're using 100W+ on the CPU isn't bad at all. It's only bad if the core temp difference gets -worse- over time.
     
    Vistar Shook likes this.
  8. Papusan

    Papusan BGABOOKS = That sucks!! STAHP! Dont buy FILTH...

    Reputations:
    10,489
    Messages:
    17,635
    Likes Received:
    25,517
    Trophy Points:
    931
    5°C above other cores is only bad if the cooling can't keep it cool whatever clock speed you want to run. Some on the forum here advice everybody... You don't need to fiddle with applying new or better thermal paste. Stock thermal paste is more than good enough. As long it doesn't run into throttling teritory or 93°C there is NO problem. Aka you should do nothing, or better say no need for it. What if you later want to increase clocks above the clocks you had when testing? Or take with you your notebook out other places with much higher ambient temp?
     
    Last edited: Apr 16, 2018
    Darkhan, Falkentyne and Vistar Shook like this.
Loading...

Share This Page