*** The Official MSI GT75 Owners and Discussions Lounge ***

Discussion in 'MSI Reviews & Owners' Lounges' started by Phoenix, Jun 23, 2017.

  1. Falkentyne

    Falkentyne Notebook Prophet

    Reputations:
    4,582
    Messages:
    4,063
    Likes Received:
    5,378
    Trophy Points:
    581
    Hi tweety78. The original reason I said to not use static voltages is because whatever you put for static voltages will be automatically boosted by 100-150mv at full load due to the default IA AC DC setting. So the voltage you put in for static is nothing close to what you are getting, making this option completely worthless and misleading.

    When you set IA AC DC to "1" then the MSI VID boost shenanigans is disabled, so then your static voltage functions properly. MSI also does 'voltage droop compensation' by an internal setting which cannot be disabled. This is why the "DC" setting has to be set to 1, rather than 179 or 180 (which is Auto/default). The IA DC setting tells the Intel CPU how much to *REPORT* to the system the droop in the VID, based on current going to the chip--note that this setting does NOT actually droop the real voltage! that is handled by the onboard voltage regulators; the Intel designed specification for vdroop is dropping the voltage by a factor of 1.80 mOhms at full load. This specification on desktop systems can be overridden by something called "Loadline Calibration", which intercepts the CPU VCORE directly, rather than the VID).. The higher the DC setting, the more the VID will drop at full load. The actual VID going into the CPU itself is controlled by the IA AC Setting. How much VID is reported (after the IA AC setting is fed into the CPU) back to the operating system is based on the IA DC Setting (and some other factors).

    On a Clevo or eVGA system, setting IA AC loadline to 1 and IA DC loadline to auto (or 180; 210 for coffee lake LGA systems) will prevent the voltage from being boosted at full load, and will show the VID to drop by a factor of 1.80 mOhms at full load, which should correspond closely to the voltage the CPU is getting. However MSI systems actually internally "remove" the intel designed voltage droop, which means the IA DC setting must also be set to 1. If this makes sense.

    In a twisted way, this is another reason why if IA AC DC Are both set to 179 (default), why at full sustained load (continuous load), the VID Shown will be much lower than the actual VID going into the CPU, because the IA DC setting will droop the VID by 1.80 mOhms of current resistance, but MSI's internal "Loadline Calibration" has already removed that droop to begin with, causing the VID to be misreported as too low.

    You should google what Loadline Calibration actually is first.
    Note: Loadline Calibration (LLC) functions DIRECTLY on CPU Vcore.
    IA AC and IA DC loadline functions directly on CPU VID.
    VID is not vcore, however depending on the design of the laptop or desktop, VID can heavily influence vcore, and can be very accurate in relation to vcore, or very very inaccurate.

    The really high VID spikes you see with IA AC DC=auto, are actually done at very light load, due to fluctuations in the voltage regulation, where IA AC will boost the VID far up, and IA DC won't report any droop at all.

    Ok you want the TL;DR version right?

    Ok. If you REALLY want to see exactly how much your MSI laptop is boosting and falsely reporting your VID , set IA AC loadline to 179 (or auto) and set IA DC loadline to 1. This will cause the full load sustained VID to be accurately reported (not 100% but very close).

    Then watch what happens at full idle, then at full load (this will be no different than IA AC DC=auto, except THIS time, the VID will rise at full load. Prepare to have your mind blown).

    Now if you want to see a grossly horribly inaccurate VID, try this:
    Set IA AC loadline to "1" and set IA DC loadline to 400.

    Watch what happens to your CPU VID at full load. Just watch.

    Have fun.
    (note: for some strange reason, heat and power draw are slightly higher at IA AC=1, DC=400, rather than IA AC=1, DC=1. Just please don't try something utterly stupid like IA AC=400, DC=1, please).
     
    Last edited: Jul 8, 2018
    raz8020 likes this.
  2. Papusan

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

    Reputations:
    12,413
    Messages:
    18,535
    Likes Received:
    27,949
    Trophy Points:
    931
    Shouldn’t it be 210 ?
     
  3. Talamier

    Talamier Notebook Enthusiast

    Reputations:
    22
    Messages:
    11
    Likes Received:
    17
    Trophy Points:
    6
    Ok wow this helped immensely. Thank you for the recommendation! It's mildly frustrating all the tweaks you have to do these days just to get full performance. My old Win7 tower kicked ass the day I built it and turned it on. I'm driving a 3440x1440 ultra wide monitor so I need to squeeze as much performance as I can. My thermals never seem to exceed 76C thanks to the HIDevolution re-pasting so I may dabble in some GPU overclocking in the future.
     
    heliada and Donald@HIDevolution like this.
  4. Donald@HIDevolution

    Donald@HIDevolution Company Representative

    Reputations:
    10,324
    Messages:
    8,797
    Likes Received:
    4,398
    Trophy Points:
    431
    I am so glad you are pleased!

    Squeezing this much horsepower into a little box does take good tuning. Our "Secret Sauce" to tame the temps involves more than just the Grizzly Conductonaut + Fujipoly Extreme Thermal Pads and Bottom Ventilation Mod upgrades. The other blend of ingredients comes standard when you get the Thermal Material upgrades.
     
    Talamier and Phoenix like this.
  5. heliada

    heliada Notebook Evangelist

    Reputations:
    152
    Messages:
    376
    Likes Received:
    363
    Trophy Points:
    76
    You are welcome ^^ I wish I knew about hidevolution a year ago when I bough my gt72vr from a local shop... Now I am all on my own with troubleshooting aside from the extremely lovely people on these forums! But since even Phoenix has issues with the new drivers, I think everyone is affected (so am I). I raised the issue with msi and nvidia but nothing is happening as I am just an individual. It's sad. Older drivers give me bsod & black screen issues when the display turns dark to save power as per windows power settings - if I disable that, no issues. All the drivers available for this laptop are a mess imo and I wish nvidia or msi did something but I can't do anything beyond submitting my feedback to them.
     
  6. Falkentyne

    Falkentyne Notebook Prophet

    Reputations:
    4,582
    Messages:
    4,063
    Likes Received:
    5,378
    Trophy Points:
    581
    Yeah but MSI sets it to 179 by default. The reference value for KABY LAKE is supposed to be 1.80 mOhms and for Coffee Lake 2.10 mOhms. But this is for LGA processors on desktop boards. But we're dealing with BGA cancer here. Now it's possible the referenced Coffee Lake value over on the Asus forums is a mistake by Raja (the Asus employee), after all, 8700K can be used on Kaby lake boards with modification, but someone needs to find the intel technical documents. In fact you know Intel DOES have an 800 number. Someone could call them and ask for engineering reference documents for the IA Loadlines. Won't be me, however. I already put in my time.
     
    Papusan and heliada like this.
  7. Tweety78

    Tweety78 Notebook Guru

    Reputations:
    34
    Messages:
    58
    Likes Received:
    59
    Trophy Points:
    26
    Thanks for the detailed explaination, I think though I will stay with AC/DC set to 1 and not mess with it.
    When HIDEvolution support walked me through changing IA AC/DC loadline to 1 in the BIOS, he also had me change to static voltage of 1.250v, that's what you see in the screenshot. who knows, maybe MSI will correct the AC/DC default values in the future so customers wouldn't have to fix it on their own.
     
  8. Falkentyne

    Falkentyne Notebook Prophet

    Reputations:
    4,582
    Messages:
    4,063
    Likes Received:
    5,378
    Trophy Points:
    581
    The default values are designed specifically by Intel. MSI has no say in this.
    But these values are designed to work with ADAPTIVE voltages. It's made for automatic adaptive voltages.

    On desktop boards, when using STATIC voltages (override voltages), this setting is ignored, because static voltages directly affect CPU VCORE rather than VID and the IA AC DC setting thus gets ignored,, so the IA AC DC loadline setting does not affect the manual vcore setting.

    the issue with MSI isn't even this.
    It's that they use loadline calibration to compensate for vdroop, when the IA AC DC setting basically already does its own "vdroop compensation" itself by boosting the VID by a certain amount of resistance at full load (IA AC), and then having it droop back down to prevent overshoot (IA DC). On desktop boards, "Loadline Calibration" is used to compensate for vdroop, and usually LLC is disabled when using fully adaptive voltages. so you can see where the problem can occur. You basically have the Intel designed boost/droop setting doing its own voltage droop/compensation, then MSI's own loadline calibration setting also doing it, so then the IA AC setting boosts an already flatlined voltage curve, causing it to 'Jump' at full load.

    This is why for example, Vistar Shook's 7820HK on an eVGA laptop needs 1.50v set in the Bios to do a 4.9 ghz cinebench run, while MSI' (my sample) needs 1.38v to do a 4.9 ghz cinebench run, yet his temps and my temps and power draw on both chips is basically identical. His eVGA doesn't have an internal loadline calibration, so his 1.50v voltage gets drooped down ant load (probably to around 1.35-1.40v), AND his eVGA also ignores the IA AC DC =auto setting when using STATIC VOLTAGES (it basically acts like it's set to 1 and 1 already), while MSI uses 1.38v and auto compensates for the vdroop (AND also needs IA AC DC loadline set to 1 manually to prevent voltage BOOST).
     
    Tweety78 likes this.
  9. NerdAlert

    NerdAlert Newbie

    Reputations:
    5
    Messages:
    9
    Likes Received:
    1
    Trophy Points:
    6
    Has anyone been able to locate a detailed GT75 Titan 8RG dissasembly guide? I've searched everywhere and can't find what I'm looking for... I would think there has to be something out there that details how everything is assembled. I've seen some pretty crappy videos and I understand how one would perform a board flip, but I'm looking for other minor things. Right now for instance, all I'm looking to do is replace the right button on the touchpad. I dont feel comfortable trying to pry it out and would like to know exactly how it is connected if I can avoid a full teardown for such a small replacement.
     
  10. Phoenix

    Phoenix Super Tweaker

    Reputations:
    13,519
    Messages:
    17,498
    Likes Received:
    24,111
    Trophy Points:
    931
    GT73VR Titan Pro disassembly video but they should be very similar:



     
    Vistar Shook and Papusan like this.
Loading...

Share This Page