Previous 1 22 23 24 25 26 Next 2,269 Replies Latest reply: Feb 21, 2015 4:45 AM by oGr3 Go to original post Branched to a new discussion.
  • John Harrold Level 1 Level 1 (5 points)
    Thermal paste is used to increase conduction. Basically when you have a heat source (cpu) with a heat sink on top of it the thin gap air that lies between the two creates a huge resistance to heat transfer. The thermal paste basically fills this void, makes it easier for heat to flow from the cpu to the heat sink. I don't really see how excess paste will result in any significant increase in heat buildup with the cpu.

    I understand you should be weary of what you read on message boards and whatnot, but I'm not just pulling this out of my rear end. I have a PhD in chemical engineering and transport (both mass and energy) is a pretty fundamental aspect of this field.
  • Chullo Level 1 Level 1 (0 points)
    Well you are the expert lol. Umm, quick question though, wouldn't increasing the contact area have a negative effect when it comes to thermal paste? It's always been known to me that a super conductive thermal paste like arctic silver for example will result in too much heat being retained in the CPU. It's basically in the instruction manual. I'm thinking that too much thermal paste results in a thicker than normal layer being present above the CPU, reducing heat transfer. I don't know the chemical side of the equation though.
  • Schwa72 Level 1 Level 1 (15 points)
    John Harrold wrote:
    Thermal paste is used to increase conduction. Basically when you have a heat source (cpu) with a heat sink on top of it the thin gap air that lies between the two creates a huge resistance to heat transfer. The thermal paste basically fills this void, makes it easier for heat to flow from the cpu to the heat sink. I don't really see how excess paste will result in any significant increase in heat buildup with the cpu.

    I understand you should be weary of what you read on message boards and whatnot, but I'm not just pulling this out of my rear end. I have a PhD in chemical engineering and transport (both mass and energy) is a pretty fundamental aspect of this field.

    John, I just assumed that metal-to-metal contact was an inherently more efficient way to transfer heat than using thermal paste between the two metal surfaces. If that's the case, then I'd think that too much thermal paste would reduce metal-to-metal contact and thereby reduce heat transfer.

    FWIW, I had one heat-and-mass transfer class during undergrad but that was a loooonng time ago.
  • Horniasty Level 1 Level 1 (0 points)
    My situation is also not heat related at all. As I described in short what I experience (page 20)
    I don't need any test, any stress put on the system, on any number of cores - all I need is change my "i" to "d" on the gfxCardStatus to make it all stop, instantly, no questions asked, no given time for any answer..

    by the way, if not the gfxCS I would simply put that metal box in that cardboard box back, not really knowing what to do with a flat LED lamp with some keys on the bottom.

    the Intel graphics didn't produce ANY freeze and any problem; I did try to test it as well to check, all stable
  • jkirker Level 1 Level 1 (0 points)
    Adrian, I've been playing with the 2.2 machine all day and have had freezes as quick as 10-15 seconds and as far out as 21 minutes, 49 seconds.

    I'm using Temperature Monitor by Marcel Bresink and it's showing individual temps for 8 cores. With no load all of my cores are hanging out at about 53c/54c with the exception of Core 4 which is showing 82c. Under load however, all w/ SmallLuxGPU all of them get up to 96c-98c before failure.

    If Adrian's machine is topping out at 90c he's in a much better position.

    BTW - I did just hang up the phone with a senior advisor and she said that engineering is working on a fix however there is no ETA. I'm still replacing this beast though with a new box that will hopefully behave.

    Obviously, they are monitoring this thread so keep posting system info, symptoms and things that are causing and helping the situation.
  • lithast Level 1 Level 1 (0 points)
    A couple of observations from testing it under bootcamp;

    EDIT: Retracting previous observations, the program I was using was displaying incorrect data. Turbo boost working as intended.

    Message was edited by: lithast
  • rockngroll Level 1 Level 1 (0 points)
    I tried it and you're right. Using the integrated graphic chip, everything works perfect. Rendering, benchmark. Hmm, so maybe a problem a firmware update could solve? Or should I sent it back? Maybe there is a chance to get a new and working one? What do you think?
  • Chullo Level 1 Level 1 (0 points)
    What benchmarks are you using and what temps is your core getting to?
  • John Harrold Level 1 Level 1 (5 points)
    It's always been known to me that a super conductive thermal paste like arctic silver for example will result in too much heat being retained in the CPU.


    Energy flows from regions of high temperature to low temperature. If you stick thermal paste on a cpu that is generating heat, the thermal paste isn't going to suck up the heat and not release it if there is a heat sink sitting on the other side of the thermal paste that is cooler. One would expect the heat sink to be cooler because it is going to have a side with fins (to increase the surface area) that is exposed to the air. The only way heat is not going to flow from the cpu -> thermal paste -> heat sink -> air is if the air is at the same or greater temperature as the heat sink. This is basically the second law of thermodynamics.

    John, I just assumed that metal-to-metal contact was an inherently more efficient way to transfer heat...


    In theory metal to metal is good because metal is generally a good conductor of heat. If this was sufficient then no thermal paste would be used at all. However, you cannot make two pieces of metal perfectly smooth. No matter how hard you try you'll have a small gap between the two. +Note this is the same principle of double pane windows in a house where you want a gap of stagnant air to prevent heat from leaving in the winter and entering in the summer+. When the thermal paste is applied most heat sinks are then placed on top of the cpu and some form of pressure is used to hold them on. The thermal paste normally isn't viscous enough to provide resistance and hold the two pieces of metal apart (cpu and heat sink). So basically the thermal past just fills in the gaps.

    The pressure of the heat sink will force any excess thermal paste out along the edges. It is possible that this can squirt out the side and then somehow get on top of the heat sink (this is the only case were I can see it causing problems) and prevent heat from leaving due to convective heat transfer along the surface -- it fills in the fins on the surface, reducing the surface area and the rate of heat transfer to the air. We're talking a huge amount of excess here and it would probably end up slathered all along the inside of the laptop as well.
  • Chullo Level 1 Level 1 (0 points)
    Alright John, I'll respectfully disagree on that. It's just a known quantity in the computer assembly process that you do not use excess thermal paste. It was the same issue apple had with their 1st gen MacBook. It's what we in the repair business know 1st hand, i'm not particularly knowledgeable about the exact mechanism, but it's fact. A very thin film of thermal paste is the required amount. Not a slightly thicker one, and these are very minute differences. Couple that with a very high horsepower processor and the same cooling system from a CPU that was 10W (35 vs 45) less power hungry and you have problems.

    Another thought I had was whether they may have had a bad batch of thermal paste. This would have affected a very large swath of machines (as is evidenced here).

    Message was edited by: Chullo
  • John Harrold Level 1 Level 1 (5 points)
    I suppose we will have to agree to disagree on this one. I didn't want you focusing on something that was inconsequential while trying to debug this problem.

    By the way, as an undergrad I worked as a computer technician doing computer assembly and repair
  • rockngroll Level 1 Level 1 (0 points)
    I tried everything I did before, like Video rendering under iMovie and Premiere CS5 - no crash with integrated card on. I also did the smalluxgpu Test with cpu*8+GPU and 800x600 resolution. With ATI Card on, the System freeze after a 10 seconds on 91 C°. With Intel Graphic I've stopped the test after 300 seconds. Temp between 89 and 94 C°.

    Temp idle with safari on and integrated card: 50 C°

    Message was edited by: rockngroll
  • Chullo Level 1 Level 1 (0 points)
    John ----- Nice. Doing it now as a natural talent while I do my masters. What do you think of my last thought? ....whether it may have been a bad batch of thermal paste?

    Message was edited by: Chullo
  • Chullo Level 1 Level 1 (0 points)
    Rockngroll --- you my friend are in the same boat as most of us here....heating issue probably combined with some other driver/firmware problem. The point is, the issue reproduces itself when our temps go up and approach 100C.
  • Schwa72 Level 1 Level 1 (15 points)
    John, I understand that thermal paste basically "fills in the gaps" between the metal-to-metal CPU-to-heat sink contact. My assumption is that if too much thermal paste is applied, not all of it would "squirt" out when the heatsink is clamped to the CPU and consequently very little, if any, metal-to-metal contact would remain.

    I will tell you that in enthusiast computing circles it's recommended that the minimum effective amount of thermal paste be used. In my personal experience building gaming PCs, there's not very much clamping force between the heat sink and the CPU. Since thermal paste is pretty viscous, it's feasible that too much thermal paste could inhibit metal-to-metal contact between the CPU and the heatsink simply because the excess paste wouldn't "squirt" out the side.

    In my opinion it's inaccurate to assume that all excess thermal paste, regardless of how much is initially applied, will "squirt out" once the CPU and heat sink are clamped together. If you use too much paste, some will remain in the CPU-heatsink gap, effectively reducing heat conductivity by reducing the metal-to-metal contact area.

    Since we're throwing around academic credentials, I'll mention that I have an MS in aerospace engineering with a focus in computational fluid dynamics, so I know a little about fluid viscosity and the like.
Previous 1 22 23 24 25 26 Next