3 ms·

Coils emit a relatively high-frequency sound, so your best bet is a case with thick panels, that’s fairly well enclosed, and has some acoustic dampening. Most

by ovao 8y ago

Coils emit a relatively high-frequency sound, so your best bet is a case with thick panels, that’s fairly well enclosed, and has some acoustic dampening. Most cases made today don’t have any significant dampening material. It’s pretty trivial to add some to the panels without significantly affecting cooling capacity.

cellularmitosis 8y ago
Alternatively, if you wanted to really dig into this, you could try to find every switch-mode power supply and replace them with (much less efficient) linear power supplies. You'd need a larger water-cooling system though.
- hatsunearu 8y ago
  Is this a joke? That would require making your own GPU PCB, and maybe a year's worth of studying on electronics and power supply design.
  - cellularmitosis 8y ago
    Not really. More likely just desoldering a couple of key components and soldering in a few wires to a dsughterboard with a linear regulator. A year’s worth of power supply design? Have you looked at the LM7805 datasheet? The circuit is one regulator and two capacitors...
    - magicalhippo 8y ago
      Have you looked at the current requirements for CPUs and GPUs? We're talking 100A or more. Even if the input was 3.3V, which it isn't, a linear regulator down to 1.1V would have to dissipate 220W or more. You'll need a bigger heatsink for the linear regulator than for the CPU... And designing a 250W-capable linear regulator is not as simple as just hooking up a LM7805. So with a modern CPU and GPU, your talking ~400W of power to the actual components, and nearly 500W wasted in the linear regulators. This of course also means you have to get a 1000W PSU as a bare minimum.
      - hatsunearu 8y ago
        Here's the problem: not only do you have to dissipate 200W+ for your pass element, you need to drive it ultra fast with an extremely fast analog circuit that can withstand the massive magnetic fields (which pretty much means you need a PCB). Yeah and good luck driving a 200W linear element (if it even exists, lol) with a few op amps--the driver which should deliver a few amps into the gate/base of the pass element, which in and of itself is a pretty difficult challenge.
    - hatsunearu 8y ago
      old thread but LMFAO you can't be any more wrong. You need /much/ more careful design to get GPU-compliant performance. The dI/dt on modern ASICs are insane, and you need an insane regulator to deal with it.
lutorm 8y ago
I was going to say, I can hear coil whine if I stick my ear into the machine, but it doesn't make it out past the acoustic dampening panels in my case.

by ovao 8y ago

cellularmitosis 8y ago

Alternatively, if you wanted to really dig into this, you could try to find every switch-mode power supply and replace them with (much less efficient) linear power supplies. You'd need a larger water-cooling system though.

hatsunearu 8y ago
Is this a joke? That would require making your own GPU PCB, and maybe a year's worth of studying on electronics and power supply design.
- cellularmitosis 8y ago
  Not really. More likely just desoldering a couple of key components and soldering in a few wires to a dsughterboard with a linear regulator. A year’s worth of power supply design? Have you looked at the LM7805 datasheet? The circuit is one regulator and two capacitors...
  - magicalhippo 8y ago
    Have you looked at the current requirements for CPUs and GPUs? We're talking 100A or more. Even if the input was 3.3V, which it isn't, a linear regulator down to 1.1V would have to dissipate 220W or more. You'll need a bigger heatsink for the linear regulator than for the CPU... And designing a 250W-capable linear regulator is not as simple as just hooking up a LM7805. So with a modern CPU and GPU, your talking ~400W of power to the actual components, and nearly 500W wasted in the linear regulators. This of course also means you have to get a 1000W PSU as a bare minimum.
    - hatsunearu 8y ago
      Here's the problem: not only do you have to dissipate 200W+ for your pass element, you need to drive it ultra fast with an extremely fast analog circuit that can withstand the massive magnetic fields (which pretty much means you need a PCB). Yeah and good luck driving a 200W linear element (if it even exists, lol) with a few op amps--the driver which should deliver a few amps into the gate/base of the pass element, which in and of itself is a pretty difficult challenge.
  - hatsunearu 8y ago
    old thread but LMFAO you can't be any more wrong. You need /much/ more careful design to get GPU-compliant performance. The dI/dt on modern ASICs are insane, and you need an insane regulator to deal with it.

lutorm 8y ago

I was going to say, I can hear coil whine if I stick my ear into the machine, but it doesn't make it out past the acoustic dampening panels in my case.