The BookShelf Server

[Lost-Benji]

Now I know it is for the 120mm fans but it does impart a great deal of knowledge.
Noctua.at - sound-optimised premium components "Designed in Austria"!
Take note of the F12 over the other two.

Noctua.at - sound-optimised premium components "Designed in Austria"!
See how it has guide vanes to help with ensuring high static pressure. These are the fans to use when pulling through filters and RADs.

Don't get too caught up in volumes of airflow, pay more attention to the noise and speed of which they spin to produce their flow. Faster means more noise in most cases.

I cant help you with a inline power meter, I live in Australia and we run decent voltage/currents compared to chicken-countries (240VAC @ 10A Vs. 110VAC @ 15A)

 

[PigLover]

"Enough" airflow depends on a lot of variables, including the volume of the space you are managing and the static pressures involved. You've got a lot of volume and don't have anything major in the form of obstructions (i.e., your not trying to pull the air through a group of very tightly places disk drives in a rackmount). Plus you've put a good deal of thought into where the sources and exhausts are and how the equipment lines up in between. This actually makes it a pretty easy space to cool

My intuition says yes, you'll be OK with these four fans. But that's just intuition, not any kind of engineering. Frankly, I'd place them all at the exhaust side of your cabinet and let them "pull" air through it. This is what is known as a "negative pressure design". Others prefer the positive pressure approach of putting the fans on the source side. In truth both approaches are valid and have trade-offs. It is worth nothing that most of the sound-optimized fan designs (Noctua and others) are optimized for pulling air, not pushing it. Pushing air at low noise levels generally requires a blower design, not a fan.

Don't forget on the noise calcs - add 3db for each fan (if each fan generates 19.2db of noise you will get 22.2db with 2 fans, 25.2db with three, etc). A whole lot of quiet fans still make a pretty good racket when running together

Interestingly, SilentPCReview just did two roundup tests of 140mm fans. See here and here. The Noctua's you are looking at are reviewed in the first link.

 

[Kurly_B]

Thanks PigLover, great info. I think putting the fans at the top is easier for me anyway. Working on the the cable routing.

Update on the parts, just got everything, and have it booted up, running a quick funtoo install ( if there is such a thing ), to test it. Really excited just to see this board run.
Just chrooted, emerge --sync, heheheh giddy....

 

[Kurly_B]

Just set make.conf to MAKEOPTS="-j25" heheheheh

 

[Kurly_B]

I have successfully booted funtoo linux with uefi boot...... nice!!!!! custom kernel and all, no bootloader - grub or extlinux, straight UEFI.

Update:
Running board at idle, no air-conditioning running, in the original setup, this is sensors output:

localhost ~ # sensors
coretemp-isa-0000
Adapter: ISA adapter
Physical id 0: +39.0 C (high = +81.0 C, crit = +91.0 C)
Core 0: +36.0 C (high = +81.0 C, crit = +91.0 C)
Core 1: +35.0 C (high = +81.0 C, crit = +91.0 C)
Core 2: +37.0 C (high = +81.0 C, crit = +91.0 C)
Core 3: +38.0 C (high = +81.0 C, crit = +91.0 C)
Core 4: +37.0 C (high = +81.0 C, crit = +91.0 C)
Core 5: +38.0 C (high = +81.0 C, crit = +91.0 C)

coretemp-isa-0001
Adapter: ISA adapter
Physical id 1: +42.0 C (high = +81.0 C, crit = +91.0 C)
Core 0: +38.0 C (high = +81.0 C, crit = +91.0 C)
Core 1: +41.0 C (high = +81.0 C, crit = +91.0 C)
Core 2: +42.0 C (high = +81.0 C, crit = +91.0 C)
Core 3: +38.0 C (high = +81.0 C, crit = +91.0 C)
Core 4: +36.0 C (high = +81.0 C, crit = +91.0 C)
Core 5: +39.0 C (high = +81.0 C, crit = +91.0 C)

 

[gavin]

wow 7 degrees difference between cores on the same mobo. 0/Core 2 must be facing Florida or something!

 

[Lost-Benji]

I am guessing you haven't worked with 2P or 4P systems too much. It is normal to have a difference regardless of airflow or cooling. Just a tiny difference in voltages from the independent VRM's, differences in CPU's and many other variables will see variances.

 

[Kurly_B]

Hi all.
Not really made any progress. Haven't had time to work on the system. My wife had a death in the family on the same day as my daughter's first birthday , and she had to go over to the house to pronounce and sign the death certificate (she's a doctor), Been a little strange around here.

Yeah the temps that I posted are not based on the illustrated build that I posted, just bolted the thing in the original config and let it rip.....
CPU One is directly above and in the path of CPU Two's HSF discharge, so the difference in temp is expected, but those temps should change, at least I hope they will.

I'll update as soon as I can.



BTW, I need some more spinning platters, 2 to 3 TB, any recommends?
RMA's are a bitch from here, trying to avoid duds. I like to buy in lots of (4).

Going to order (4) of these this week too.

 

[Lost-Benji]

Drives depend on whats driving them. HBA's then the ST3000DM001s are cheap and work.

 

[Kurly_B]

Update - Just ordered (4) ST2000DM001, and a new TP-Link 16 port unmanaged switch, and (2) Single slot Radeon 7750's.
Haven't started the Wood Work yet. Haven't had the time. Chomping at the bit, but family drama, 1 wedding and a funeral.

Having an issue with the OS. Not sure anymore, any suggestions?
I want ZFS storage and GPU passthrough.
Played with :
Gentoo - XEN and zfs -
SmartOS - KVM no gpu passthrough
OmniOS - KVM - no gpu passthrough
XenServer - Xen - no zfs backed vm storage
Debian 7 - Xen and zfs - basically worked, but a pain.

Was thinking about ESXi, but next week ordering 32GB more ram (64 GB Total), so no go, unless I spend $500 on a license.