Which Thermal Paste to buy and apply (Traditional and Liquid Metal)

Discussion in 'Hardware Components and Aftermarket Upgrades' started by Vasudev, Jul 11, 2017.

  1. Rooter1234

    Rooter1234 Notebook Consultant

    Reputations:
    206
    Messages:
    170
    Likes Received:
    215
    Trophy Points:
    56
    Albestech tested a few top notebook pastes like SYY-157 and FuzeIce Plus on desktop CPUs (8700k and 2700x)

    https://albestech.com/review-syy-157/
    https://albestech.com/review-fuzeice-plus/

    Very good results, only the Thermalright TFX performed a bit better. IONZ IZP14 is also quite good there, it's a bit too runny for a laptop though.


    Luumi tested MX5 here:




    I think 2-2.5 degrees is quite a big difference compared to Kingpin KPx considering he is testing on a desktop CPU. Of course 2-3 degrees shouldn't matter for most but same could be said for most of the pastes, a cheap GD900 would do the job as well:

     
    Jff007, Vasudev and Falkentyne like this.
  2. stumbler

    stumbler Notebook Consultant

    Reputations:
    4
    Messages:
    150
    Likes Received:
    26
    Trophy Points:
    41
    NOTEBOOK REVIEW????
     
  3. Deks

    Deks Notebook Prophet

    Reputations:
    1,272
    Messages:
    5,201
    Likes Received:
    2,073
    Trophy Points:
    331
    I wouldn't say 2-2.5 degrees is a big difference... especially on a desktop where temperatures would/should never really reach anywhere near critical levels where such margins would be relevant. Its good for testing purposes to see which pastes have better margins though... but 2.5 degrees C is more or less just above the margin of error.

    For laptops... eh, it would be a good difference yes, but personally, I'd see 5 degrees reduction as a 'tangible' difference, and 10 degrees being 'big'.
     
  4. Rooter1234

    Rooter1234 Notebook Consultant

    Reputations:
    206
    Messages:
    170
    Likes Received:
    215
    Trophy Points:
    56
    It's the difference between a front runner and the midfield in many desktop tests because the gaps are a lot smaller for desktop chips.

    And I have to disagree with you 2.5 degrees isn't margin of error unless the test is somehow flawed.
     
    Papusan and Jff007 like this.
  5. Deks

    Deks Notebook Prophet

    Reputations:
    1,272
    Messages:
    5,201
    Likes Received:
    2,073
    Trophy Points:
    331
    I did say 2.5 degrees is 'more or less just above the margin of error'... but we're talking a difference of maybe 1.5 degrees which is statistically speaking irrelevant... so, borderline margin of error.

    Anything above that (such as say between 3 to 5 degrees) is statistically more relevant (still small, but 'noticeably tangible')... with 10 degrees being quite big of a difference.

    This is looking at things from a notebook point of view.
    I wouldn't want to tether at the edge of a system shutdown (but that also has more to do with colling designs present in laptops)... I'd prefer the chip is much cooler (about 10 to 20 degrees below that point) while fully stressed... otherwise, the intense heat from the chip would be affecting other components... and even in desktops this can result in problems down the line (what with RAM slots failing for example as they are usually placed right next to the CPU).
     
  6. Jff007

    Jff007 Notebook Enthusiast

    Reputations:
    15
    Messages:
    44
    Likes Received:
    22
    Trophy Points:
    16
    It also depends on the application. For example, in a SFF build, 2-3C is a very significant differance in temps! (As SFFPCs tend to very similar to notebooks for heat management requirements.)

    Want to add some data to the discussion as well: I repasted a desktop GPU (RTX 2080ti FE) with SYY-157, and redid my XPS 15 with SYY-157. Unfortunately, I was a bit slow and forgot to screencap before temps. Since I haven't been traveling enough to use my XPS often, I only have a note here that states the temp differences I had seen. I'll be sure to grab screenshots when I swap the XPS over to the IC Graphite.

    GPU:
    ------Stock:
    --------------Idle: 49C (Hotspot temp was 64C)
    --------------Load: 82C (Hotspot temp was 107C---at above 80C on core, fan 1 will ramp up to high RPM mode, irregardless of any custom fan settings.)
    ------SYY-157:
    --------------Idle: 40C (Hotspot temp was 53C)
    --------------Load: 74C (Hotspot temp was 89C)

    XPS 15 9560:
    Was using Cryonaut, which was now about 4 years old. A fresh application to SYY-157 saw a temp drop of 2 degrees at idle and 4 at load. So, not bad. Especially if the SYY paste pumps out less.

    I also want to note that I repasted my desktop's CPU with fresh Cryonaut (the application on it was now 2 years old). The old paste had pumped out some and was very dry. After a repaste I saw a 4C drop across the board, so Cryonaut lost about 2C in effectiveness per year of use. I'll definitely be swapping over to SYY-157 on the next repaste to test differences between the two.
     
    Last edited: Sep 11, 2021
    Rooter1234 and Atma like this.
  7. Temp1234453

    Temp1234453 Notebook Consultant

    Reputations:
    15
    Messages:
    214
    Likes Received:
    33
    Trophy Points:
    41
    You only changed that value to achieve it? How many degrees dropped? Is your CPU undervolted? Did you have to change undervolt values to keep them stable?
     
  8. Falkentyne

    Falkentyne Notebook Prophet

    Reputations:
    8,396
    Messages:
    5,992
    Likes Received:
    8,633
    Trophy Points:
    681
    That's very interesting.
    You went from an idle hotspot delta of 15C pre to 13C post, and a load hotspot delta of 25C pre to 15C post
    so idle to load delta post repaste increase was only 2C (13C to 15C) after repaste, rather than a 15C increase.

    Did you also replace thermal pads at the same time? Because on Ampere, hotspot is the hottest of either GPU *and* VRM (not VRAM) temp sensors. No idea if that applies to Turing or not. (On Pascal, hotspot is useless as it's a nonchanging fixed value).
     
    Jff007 likes this.
  9. Jff007

    Jff007 Notebook Enthusiast

    Reputations:
    15
    Messages:
    44
    Likes Received:
    22
    Trophy Points:
    16
    To be honest, I'm not sure if it pulls VRM with Turing either. I left the thermal pads exactly as they were, no new ones. It was definitely shocking to see the hotspot delta have such a dramatic difference (though the original paste job was perhaps the worst I've ever seen on any PC component.)
     
    Falkentyne likes this.
  10. Rooter1234

    Rooter1234 Notebook Consultant

    Reputations:
    206
    Messages:
    170
    Likes Received:
    215
    Trophy Points:
    56
    Another Kingpin KPx and MX5, MX4, Thermalright TFX, Noctua NT-H2 test on a Ryzen 9 5950X:


     
    Vasudev and Jff007 like this.
Loading...

Share This Page