Mini ITX build (Intel Atom powered)
-
I don't think I've read about a single failure in a PicoPSU, I stand to be corrected though. Even the most efficient ATX PSUs are still pretty bad at 5% load. Anyway so long as you were aware of it.
Edit: That PSU looks to still manage 75% efficiency at 25W though so not too bad:
http://www.tomshardware.co.uk/power-supply-psu-80-plus,review-32016-12.htmlSteve
I had one that was DOA 3 or 4 years ago. The replacement had been running 24/7 ever since.
-
As far as your PSU goes, you bought a 380w PSU.
Do you know how much power your rig draws on a regular basis?
I am just afraid if I had to buy a PSU similar to what you have (in case I go with a 1u mini-itx case and can't use the picoPSU) that the PSU is going to be drawing way more power than id like it to use. Because i have been looking at 150w,180w and 250w 1u PSUs.
-
The power rating of PSU is the maximum power it can supply not how much it draws.
A typical Atom rig will be drwaing <30W though so really get the smallest PSU you can. As we discussed above a PSU designed to supply 500W into some hardcore gaming rig is not likely to be very efficient at 30W. Indeed some PSUs are particularly poor at very low Wattages. Most are designed to give >80% efficientcy at 20% load but 20% of 500 is still 100W so at 20W it will be running at only 4%. It's not uncommon to see <60% efficiency at those load levels.
The PicoPSU is deigned for low power applications and has a very high efficiency at those levels.
Steve
-
The power rating of PSU is the maximum power it can supply not how much it draws.
A typical Atom rig will be drawing <30W though so really get the smallest PSU you can. As we discussed above a PSU designed to supply 500W into some hardcore gaming rig is not likely to be very efficient at 30W. Indeed some PSUs are particularly poor at very low Wattages. Most are designed to give >80% efficiency at 20% load but 20% of 500 is still 100W so at 20W it will be running at only 4%. It's not uncommon to see <60% efficiency at those load levels.
The PicoPSU is deigned for low power applications and has a very high efficiency at those levels.
Steve
Not only that any unused power will get translated into excess heat and efficiency hit. So really better off using PicoPSU. This will allow you to use smaller case.
-
for a mini-itx atom CPU build, would you go with a 1u case, or would you go with a small case.
also, if you go the 1u route. would you still use a picoPSU or a regular 200w or less PSU.Because, if you get the small case, then the PicoPSU would be the way to go (I think the 80w is the smallest and should work)
-
for a mini-itx atom CPU build, would you go with a 1u case, or would you go with a small case.
also, if you go the 1u route. would you still use a picoPSU or a regular 200w or less PSU.Because, if you get the small case, then the PicoPSU would be the way to go (I think the 80w is the smallest and should work)
Yep. My current setup draws under 25 watts max so it'll be perfect.
-
For those wondering, this build pulls 22.3W idle.
-
for a mini-itx atom CPU build, would you go with a 1u case, or would you go with a small case.
also, if you go the 1u route. would you still use a picoPSU or a regular 200w or less PSU.Because, if you get the small case, then the PicoPSU would be the way to go (I think the 80w is the smallest and should work)
All PicoPSU are 1u compliant.
-
Bummer, all of your pics disappeared. Any way to re-list them?
-
Are there saving to be made with PowerD? I've read mixed reviews with those using older Atoms where the CPU would be limited rather than undervolted.
Yeah, for desktop atoms that don't support SpeedStep, PowerD will use throttling instead of actually adjusting frequency and voltage, and that is actually counterproductive from an energy efficiency point of view. Even for CPUs that do support SpeedStep, it's generally best to just disable throttling by adding the following to /boot/loader.conf.local:
hint.p4tcc.0.disabled="1" hint.p4tcc.1.disabled="1"
(Add/remove similar lines as needed if you have different numbers of threads and/or cores.)
That just leaves you with the operating points that are actually useful.
-
For those wondering, this build pulls 22.3W idle.
I have a similar build, same board, 2 GiB RAM and an SSD. It's running on a PicoPSU and the power draw is 13 watts.
Now decommissioned as I'm now running pfSense virtualised under QEMU/KVM with PCI pass through on an Intel Pro/1000 VT Quad Port NIC. -
Hi all
This is my new pfSense build (the old one was a 1u Supermicro P4 box), hopefully documenting this will be useful/interesting to someone. I'm running on UK ADSL (13/0.92). Modem is a DrayTek Vigor 120 v2. A small 5 port Netgear Gigabit switch shares the connection. Wireless is supplied by a Linksys WRT54GL running the Tomato firmware.
Requirements:
- Small.
- Low power consumption (~ 25W).
- Reliable.
Parts:
Lian Li Mini-Q PC-Q07 chassis in silver.
ASRock Server motherboard AD2550R/U3S3 (Atom D2550, 2x Intel gigabit 82574L)
Kingston 2GiB DDR3 RAM
Kingston micro USB storage drive (boot drive)
Antec EarthWatts 380W PSUVarious Pics:
The router is named after the NASA capsule communicator (or capcom for short).
One gotcha when booting from a USB stick is that you need to set kern.cam.boot_delay=10000 and make the change permanent.
Hi, I'm interested in this board as well, did you perform iperf test? I want to know the actual max. throughput, from old post it seems to be ~650Mbps but that was before 2.2 (no multi cpu support) and hope to see better performance with new pfsense version