[Best Practices] Power, cooling, and general maintenace

  • Thread starter Deleted member 12106
  • Start date
D

Deleted member 12106

Guest
This is going to be a work in progress but will pot up what I've collected thus far with my experience, and the help of others on what (not) to do. Thanks grandpa, patriot, musky, and others:eek:

I do ask you please keep the chatter in thread down to a minimum. I'll update the OP as time goes on.

Cooling and power best practices for distributed computing projects

1. Choose a quality power supply
  • Power supply with higher 80 plus rating will provide lowest cost of operation, as well as produce less heat
  • From highest(most efficient) to lowest(least efficient): 80 plus platinum, 80 plus gold, 80 plus silver, 80 plus bronze, and 80 plus
  • Reputable brands such as Corsair, OCZ, or Seasonic have proven to operate at high loads for long periods of time
2. To reduce the chances of burning board connectors or melting wires:
  • Avoid EPS splitters to power motherboards
  • Avoid using PCI-E power splitters
  • If making splices always solder
  • Male PCI-E ends can be re-pinned to EPS.
3. Use duty-sized cords
  • Use cord included with PSU
  • Ensure any surge protector has a least 12ga wire when running multiple machines from one surge strip
  • Do not ever plug a surge bar into another surge bar
  • Avoid any light gauge cables, especially on high-draw applications
  • Avoid the use of extension cords
4. Cooling
  • Always ensure you have air flow over and under motherboard. Position these fans to the edge of the board. You want a good flow over top and bottom of board.
  • Ensure ambient air movement around GPUs
  • When possible, apply heat sinks to VRMs. Cool with fans
  • If not using a case (naked) do use stand offs to suspend board off of any broad surfaces
  • Use larger coolers to dissipate more heat
  • PWM fans can ramp up/down based on heat, minimum duty cycles can be specified on some motherboards. Guide on cheap PWM fans
  • Using software (windows) to manual control or monitor GPU temps or adjust GPU fan, such as ATI’s Catalyst control center, Nvidia control panel, Trixx(ATI), MSI Afterburner (ATI & Nvidia), or EVGA Precision-X (Nvidia)
  • Monitor ambient temps using a thermometer
  • IR thermometers can be used to monitor CPU sockets, boards, chips, and other devices
  • Watch temps on G34 rigs using TPC.
5. Outlets and circuits
  • Do be mindful of breaker or fusing limits. If a breaker trips on a regular basis, REDUCE THE LOAD.
  • It is recommended to use dedicated breakers for high draw applications (multiple 4p or multi GPU machines)
  • For “farm” DC projects, consult an electrician for your application. We do not advocate DIY electricians.
  • A Killawatt can be used to measure the draw in watts or amps to avoid overloading circuits.
 
Last edited by a moderator:
Good stuff here man. Lots of little things that sometimes slip the mind.
 
so I guess letting the boxen get buried in camping gear is right out
 
great guide, but just a note, TPC isn't JUST for g34 rigs, AMD in general. also, i7z for intel temp monitoring :)
 
bigted said:
great guide, but just a note, TPC isn't JUST for g34 rigs, AMD in general. also, i7z for intel temp monitoring :)
Click to expand...


Thats because sc0tty8 doesnt believe anything lower than a G34 in the AMD world to be worth mentioning....:p
 
You must log in or register to reply here.
Back
Top