Announcement

Collapse
No announcement yet.

Noctua NH-U12DX i4 + NF-F12

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #11
    Is it then maybe possible to do some cold/hot aisle separation with some plastic sheeting? It might be a relatively easy experiment that is worth doing?

    Comment


    • #12
      Originally posted by roboman2444 View Post

      Still the same amount of heat being output per each CPU, GPU, etc. The heat output doesn't suddenly increase just because you put a better fan on the machine. It will increase, however, if the components were being thermally throttled and now can run at a faster speed, or are now overclocked because of the extra thermal headroom.

      With the same amount of energy used the CPU runs cooler (=more heat is drawn away from it). Where does the heat go when it doesn't go into the air? Heat won't disappear it has to go somewhere.

      Comment


      • #13
        Ugh, this reminds what a mess the whole situation with case-mounted fans is nowadays. You have to search wide and hard for a motherboard that even has four fan sockets to begin with, then you know that half of those will not be attached to a SuperIO and always run at full speed (and thus noise) and finally those that are attached to one aren't even controllable from Linux, thanks to ACPI. Grrr.

        Why can't someone make a small, simple PCIe fan control card that would have a good amount of sockets (PWM) and a SuperIO, with the ability to turn off the ACPI access to it?

        Comment


        • #14
          Originally posted by SilverMachine View Post


          With the same amount of energy used the CPU runs cooler (=more heat is drawn away from it). Where does the heat go when it doesn't go into the air? Heat won't disappear it has to go somewhere.
          When you upgrade the heatsink, you lower the thermal resistance of the cooling system. The rate of heat dissipation (in watts) is dependent on the temperature delta and the thermal resistance. The rate of heat dissipation generally won't change at equilibrium (unless there's thermal throttling going on) as the same amount of work is performed by the CPU. Since the thermal resistance is lower for a given rate of heat dissipation, that means the temperature delta will be lower, which is why the CPU runs cooler. It's simple thermodynamics.

          Comment


          • #15
            Perhaps a(nother) portable datacenter chiller is in order, one with an intake vent that goes up to the ceiling and an output at floor level. That with the suggested heavy plastic sheeting to enclose the areas needing to be cooled would be a good idea.

            Comment


            • #16
              Michael which method of putting thermal paste do you prefer?

              Comment


              • #17
                Originally posted by GreatEmerald View Post
                Ugh, this reminds what a mess the whole situation with case-mounted fans is nowadays. You have to search wide and hard for a motherboard that even has four fan sockets to begin with, then you know that half of those will not be attached to a SuperIO and always run at full speed (and thus noise) and finally those that are attached to one aren't even controllable from Linux, thanks to ACPI. Grrr.

                Why can't someone make a small, simple PCIe fan control card that would have a good amount of sockets (PWM) and a SuperIO, with the ability to turn off the ACPI access to it?
                I don't understand the fixation on mobo fan headers and control as opposed to manaual control with constant fan RPMs. Let the freaking temps fluctuate a bit! A mobo runs fans at what RPM it pleases and software control is a hit or miss. In/out CFM ratio control is out of question as well.

                On machines with occasional changes in load you use a simple switch for 'high/low' profile. On my gaming machine I use a manual flip switch.

                ACPI? K.I.S.S.!

                Comment

                Working...
                X