That's a good point to make. I ran a new 20 amp circuit to the basement folding rack for a little bit of insurance.
 

bowlinra
I'm thinking about making another version. This one lighter weight and maybe it will move from the garage in summer to the house in winter.
 
Any thoughts on improvements or anyone else try something like this?

Might get some idea's from cp256 build for a lighter weight rack.
http://forums.evga.com/FindPost/1929306

I'm still using the original from the 1st post.. But gather up ideas and thinking..
 

That looks really great OP!
 
I have two 4x CPU boards myself and I notice the way you have setup your cpu cooling towers.
 
How did you setup the fans, are they all in one direction?
 

 
I wanted to use scenario B but I was told A would be better which how I have it setup now.
 
I assume? you are using scenario A, have you tried any other, I am unable to change the orientation of the cooling towers but I am starting to think scenario B would be better?
 
Do you have any input on this?
 
Thank you very much.
 
 

CybieC
How did you setup the fans, are they all in one direction?
 

All these pictures are my old setup, a little over year ago I redid the whole thing for 4 server rack setup and dual box fans. I never got around to posting an update.
 
I have all my setup with the fans in one direction. As well as a 20in Box every 2 servers.. My motherboard have a full 1in space between the board and wood sled that everything is mounted too. If you doing heavy OC on these with turning up the voltage, you really should air following over and under the board.

Please delete post.. images aren't coming thru properly.
 

Here is the back during the build stage.. The finished version has sides and cleaner cable management, with some screens to cover the fans and back of the computers (using for bug guards in the garage and cat guards inside the house). I'll try to get some finished pictures as well.
 

notfordman
I run one of these setups open air. In option B unfortunately the CPU's that are closest together draws the warm air extracted into the next chip. A is better only that each cpu is trying to draw in cool air. This is a very cool project.

That is exactly what I thought, my reasoning was the hot air from one CPU tower would be blown into the other rendering it ineffective.

But how about:
I still go for scenario B but like the OP I would add one of those massive fans to blow over the board
 
Would that help?
 
Or would it be better to keep A and just add one of those mega fans at the front of the board I see the OP has done the same config pretty much and if I look at the OP's pictures it seems that his 2 CPU's on the front are also rather close together like mine.
 
The reason why I really prefer scenario B is because I want to build an enclosure to protect them and I think? please correct me if I am wrong when you have Scenario A in a enclosed space it will mess with the airflow?
 
 
 
 

bowlinra
What CPUs do you have, what is your OC and what temps do you have when you run "sudo tpc -temp"?

The CPU's are currently Dual-Core AMD Opteron(tm) Processor 8216, I plan to upgrade them to 6 core.
 
Not overclocked.
 
tpc is not available for my system but heres a dump of my sensors since I installed a IPMI module :)
 
The temps seem to be OK to me.
 
k8temp-pci-00c3
Adapter: PCI adapter
Core0 Temp: +43.0°C
Core1 Temp: +36.0°C
 
k8temp-pci-00cb
Adapter: PCI adapter
Core0 Temp: +39.0°C
Core1 Temp: +42.0°C
 
k8temp-pci-00d3
Adapter: PCI adapter
Core0 Temp: +39.0°C
Core1 Temp: +38.0°C
 
k8temp-pci-00db
Adapter: PCI adapter
Core0 Temp: +37.0°C
Core1 Temp: +39.0°C
 
w83627hf-isa-0290
Adapter: ISA adapter
in0: +4.08 V (min = +0.00 V, max = +4.08 V)
in1: +4.08 V (min = +3.60 V, max = +4.08 V)
in2: +4.08 V (min = +3.60 V, max = +4.08 V)
in3: +3.04 V (min = +2.67 V, max = +3.26 V)
in4: +3.06 V (min = +2.83 V, max = +3.47 V)
in5: +3.06 V (min = +2.70 V, max = +3.30 V)
in6: +3.09 V (min = +2.70 V, max = +3.30 V)
in7: +3.30 V (min = +2.70 V, max = +3.30 V)
in8: +3.14 V (min = +2.70 V, max = +3.30 V)
Chassis Fan 1: 2445 RPM (min = 712 RPM, div = 8)
Chassis Fan 2: 2033 RPM (min = 712 RPM, div = 8)
Unused Fan: 0 RPM (min = 0 RPM, div = 2)
temp1: -48.0°C (high = +120.0°C, hyst = +115.0°C) sensor = thermistor
Northbridge Temp: -48.0°C (high = +80.0°C, hyst = +75.0°C) sensor = thermistor
Southbridge Temp: -48.0°C (high = +80.0°C, hyst = +75.0°C) sensor = thermistor
cpu0_vid: +0.000 V
beep_enable: enabled
 
w83793-i2c-1-2f
Adapter: SMBus nForce2 adapter at 2e00
VcoreA: +1.32 V (min = +1.07 V, max = +1.62 V)
VcoreB: +1.33 V (min = +1.07 V, max = +1.62 V)
in2: +0.00 V (min = +0.00 V, max = +2.05 V)
in3: +1.25 V (min = +1.07 V, max = +1.62 V)
in4: +1.25 V (min = +1.07 V, max = +1.62 V)
in5: +3.30 V (min = +2.96 V, max = +3.63 V)
in6: +0.97 V (min = +0.90 V, max = +1.10 V)
+5V: +5.07 V (min = +4.52 V, max = +5.50 V)
5VSB: +4.97 V (min = +4.52 V, max = +5.50 V)
Vbat: +3.09 V (min = +2.70 V, max = +3.30 V)
CPU 0 Fan: 2376 RPM (min = 712 RPM)
Unused Fan: 0 RPM (min = 712 RPM) ALARM
CPU 1 Fan: 2542 RPM (min = 712 RPM)
CPU 2 Fan: 2566 RPM (min = 712 RPM)
Unused Fan: 0 RPM (min = 712 RPM) ALARM
CPU 3 Fan: 2303 RPM (min = 712 RPM)
HDD Fan: 2486 RPM (min = 712 RPM)

CPU 0 Temp: +31.2°C (high = +72.0°C, hyst = +67.0°C) sensor = thermal diode
CPU 1 Temp: +34.2°C (high = +72.0°C, hyst = +67.0°C) sensor = thermal diode
CPU 2 Temp: +52.5°C (high = +72.0°C, hyst = +67.0°C) sensor = thermal diode
CPU 3 Temp: +37.8°C (high = +72.0°C, hyst = +67.0°C) sensor = thermal diode

temp5: +45.0°C (high = +80.0°C, hyst = +75.0°C) sensor = thermistor
cpu0_vid: +1.175 V
intrusion0: OK
beep_enable: disabled

bowlinra
I don't really know if the smallers are mess up the air flow. I tend to think of the rack as a wind tunnel, I want the air to flow as clean as possible from the front thru the rack and out the rear. 

Right that is what I would like since all rack servers are designed that way.
 
The thing I wonder is that you have a big fan as intake front to back while the CPU towers and side to side of that makes sense.
 
The way I reason is that if you have a big fan blowing one side and the smaller fans blowing the other side
Like this and it looks like this is exactly what the OP is doing that the blue fans would be fighting for air, but I guess if the OP has had it like this for a while now in operation it should not be an issue.
 

 
I guess I just have to give it a try for myself and test out Scenario A and B and see how much it effects the cooling.
 
It would be nice if I could have the air flow from one side to another to also cool the board and then I can make a simple wooden box with intake fans and a exhaust, I notice the chip on the SIMSO AOC IPMI module I added gets pretty warm, not to mention that the entire room they are running is nice and warm lol so these guys do generate a nice amount of heat.
 
 

Sorry. On my phone. Correct I would not be worried.

CybieC
CPU 0 Temp: +31.2°C (high = +72.0°C, hyst = +67.0°C) sensor = thermal diode
CPU 1 Temp: +34.2°C (high = +72.0°C, hyst = +67.0°C) sensor = thermal diode
CPU 2 Temp: +52.5°C (high = +72.0°C, hyst = +67.0°C) sensor = thermal diode
CPU 3 Temp: +37.8°C (high = +72.0°C, hyst = +67.0°C) sensor = thermal diode

 
I missed this earlier.. If CPU 2 isn't running some special load that's not on the other cpus. I'd say you might want to make sure that the CPU Heat sink is working properly, either the fan or the sink is loose, or the thermal paste is too thick or drying up.
 
There is a remote chance the CPU / motherboard temp sensor is broken too.

bowlinra
 
I missed this earlier.. If CPU 2 isn't running some special load that's not on the other cpus. I'd say you might want to make sure that the CPU Heat sink is working properly, either the fan or the sink is loose, or the thermal paste is too thick or drying up.
 
There is a remote chance the CPU / motherboard temp sensor is broken too.

I am guessing the sensor is dodgy, the cores themselves indicate a much lower temp, just to make sure I am just going to apply some new paste and refit that one.
 
 

It's worse than that - 24/7 operation you need to DERATE the circuit and the outlet 20%, so that's only 12 amps.
On the other hand, most areas supply at least 110 volts at the outlet, so you have about 1320 watts to work with SAFELY from a single 15 amp circuit - but make sure nothing else is on the same circuit as that outlet before you assume the OUTLET has the full power available to it.
 
 

Take a standard 20" x 20" furnace-type air filter, stick it on the intake side of the fans, and NO DUST (well, a little might manage to sneak in from the open sides, but more likely it will get blown OUT by the fan airflow).
A standard Lasko/Galax/etc box fan generates enough pressure to hold the filter in place if they are running at high, not sure about if it will work if you run them at medium or low.
Might have to mount a small "L" bracket to the bottom side of the fan on one of the "grill mount screws" to keep the filter positioned, or 2 if you have a "2 screws" side of the fan grill at the bottom.
 
 

• I really wanted the rack on UPS, but I could easily run >560W per rig. To use a single UPS, I would have had to step up to a 2200 VA system, requiring special 20A circuits, plugs and I'm not sure they even make a 20A extension cord. I really didn't look into that much, probably so have spent more time on it.

 I'm not sure about 20 amp extension cords (I THINK they exist intended for portable A/C unit usage or "job site" usage), but I have a Tripplite 20 amp 6-outlet "strip" with a 15' cord, bought it from NewEgg a few months back.