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

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

  1. prodj

    prodj Notebook Consultant

    Reputations:
    15
    Messages:
    103
    Likes Received:
    24
    Trophy Points:
    31
    Well, I'm not even dreaming of 68°C, but I have couple of questions:
    1. What test in prime95 do you run? I run the first one (Small FFTs, max heat, stress FPU).
    2. Do you think it's possible to achieve 0-1° differential between the cores with Nano paste and Arctic pads, without sanding the heat sink? (don't wanna go liquid metal and don't wanna mess with the HS).
     
  2. Falkentyne

    Falkentyne Notebook Prophet

    Reputations:
    7,078
    Messages:
    5,389
    Likes Received:
    7,444
    Trophy Points:
    681
    Small FFT with AVX/FMA3 disabled in local.txt (read the undoc.txt file)
    AVX enabled at 4.5 ghz only if ambient temps are cold (air conditioned) and I want to make fully sure there is good temp contact and its best not to run it for long past 80C. High stress temps can increase temp differentials if you exceed 80C for too long. Trying AVX small FFT with more than 1.20v real static voltage at >4.5 ghz (IA AC DC loadline=1 must be set) will just trip the Zener diodes and the laptop will shut off, because the Zeners will think that you would reach 100C with that type of current unavoidably (they are calibrated for keeping under 100C with good aftermarket thermal paste, not liquid metal).

    The highest I can run prime95 small FFT with AVX disabled is 4.7 ghz@1.276v static voltage (IA AC DC loadline=1 must be set in the bios). It's hard to keep the temps down. Small FFT AVX enabled will *instantly* trip the diodes and the laptop will shut off instantly.

    4.8ghz @ 1.310v will pass Stockfish 8 thread chess engine easily in a cool room. Prime with AVX disabled will pass 94C and the zener diodes will trip.

    4.9ghz @ 1.380v will PUBG and other games fine with temps reaching 80C, but a few runs of cinebench R15 and the temps reach 94C and the zeners trip again.

    And no its impossible. You are using direct die contact without an integrated heat spreader to touch the entire die and spread the heat evenly around for you for a large heatsink base to pick up. With direct die contact, ANY difference in heatsink pressure, thermal paste amount, even by a few microns will (or whatever the measurement of pressure is..micro PSI's?) will cause core temp differences because there is no heat spreader to help you. That's why a convex or concave heatsink is very bad (very slightly concave would be better than very slightly convex because the CPU silicon housing itself is slightly convex. So you are never going to get the temps perfect.

    With an unsanded heatsink and stretched out, compressed 0.5mm pads and a good THICK PASTE, you would be very lucky to get the temps within 3C max delta.

    And the paste will always pump out/dry out prematurely on the hotter cores side (usually by the VRM's) just because of the reduced heatsink pressure+air gaps, etc. If you can get the core temps down to 2C with good thermal paste, this shouldn't happen, but that's VERY hard to do with a convex heatsink.

    This was 4.7 ghz 1.276v in a very cold Starbucks.

    4700_82c.jpg
     
    Last edited: Aug 14, 2018
    raz8020 and prodj like this.
  3. prodj

    prodj Notebook Consultant

    Reputations:
    15
    Messages:
    103
    Likes Received:
    24
    Trophy Points:
    31
    ...
    Well, then I'll call 6°C difference good enough and leave IC graphite pad there, at least it won't dry out.
    It's really annoying to dismantle ALL the heatsinks just to get to CPU.
     
    Pedro69 and hmscott like this.
  4. maxsilver

    maxsilver Notebook Enthusiast

    Reputations:
    20
    Messages:
    20
    Likes Received:
    18
    Trophy Points:
    6
    I hear you that it is indeed a pain in the ass, but it's not THAT big of a deal all things considered. It's only 8 extra screws to remove. I gained a tremendous fan speed/temperature reduction by repasting my GPU, so at least in my case it was well worth it to repaste the GPU along with the CPU. Once you've done this more than once it's really easy to do. It sucks to waste thermal paste but after seeing some of the necessary steps to disassemble other laptops heatsinks this is really very simple in comparison.
     
  5. Pedro69

    Pedro69 Notebook Evangelist

    Reputations:
    56
    Messages:
    472
    Likes Received:
    199
    Trophy Points:
    56
    How much temperature difference you have in cpu cores? Only between max values in hwinfo.
     
  6. maxsilver

    maxsilver Notebook Enthusiast

    Reputations:
    20
    Messages:
    20
    Likes Received:
    18
    Trophy Points:
    6
    The difference between the hottest and coolest core was 6 degrees after a 5 minute Prime 95 run.
     
  7. Pedro69

    Pedro69 Notebook Evangelist

    Reputations:
    56
    Messages:
    472
    Likes Received:
    199
    Trophy Points:
    56
    Nice, i have 15ºC more in Core 2 than the others after a week, my laptop is always on anyways...
     
  8. BlakLanner

    BlakLanner Notebook Guru

    Reputations:
    10
    Messages:
    51
    Likes Received:
    28
    Trophy Points:
    26
    15 degrees? Something is very wrong there.
     
  9. Pedro69

    Pedro69 Notebook Evangelist

    Reputations:
    56
    Messages:
    472
    Likes Received:
    199
    Trophy Points:
    56
    Well, in first days its all ok, but then after a week i get that mesures...i only use games to get the values and the fans are always in Auto now.
     
  10. Falkentyne

    Falkentyne Notebook Prophet

    Reputations:
    7,078
    Messages:
    5,389
    Likes Received:
    7,444
    Trophy Points:
    681
    I explained why this happens many many many times now. I explained EXACTLY why this happens.
    And the problem can NOT be fixed without sanding. Period.
     
    Ghost 350 likes this.
Loading...

Share This Page