News: This forum is now permanently frozen.
Pages: [1]
Topic: Has Anyone used the new APU1C by PC Engines and booted via SD  (Read 5745 times)
« on: March 22, 2014, 20:42:23 »
rlgjr562 *
Posts: 3

Has Anyone used the new APU1C by PC Engines?  It is by the company that makes the ALIX series.  They say it can run m0n0wall.  But it does not directly support compact flash and instead supports booting from an SD chip.  I am not sure how to get a m0n0wall boot image on a SD chip.
« Last Edit: March 22, 2014, 20:43:55 by rlgjr562 »
« Reply #1 on: March 22, 2014, 20:52:48 »
Fred Grayson *****
Posts: 994

Do you have a memory card reader for your PC that accepts SD?

--
Google is your friend and Bob's your uncle.
« Reply #2 on: March 22, 2014, 20:56:38 »
rlgjr562 *
Posts: 3

Yes,  I have several PC's that have SD reader/writers.
« Reply #3 on: March 22, 2014, 21:53:54 »
Fred Grayson *****
Posts: 994

Then all you have to do is follow the installation instructions at:

http://m0n0.ch/wall/installation_generic.php


--
Google is your friend and Bob's your uncle.
« Reply #4 on: March 25, 2014, 13:27:12 »
brushedmoss ****
Posts: 446

I have one of these, I booted from USB stick ...

I installed an mSATA internally and installed to it from the USB stick, as i didn't want to keep booting from an external USB.

its a great board btw, got 800mbps through it at ~80% cpu.
« Last Edit: March 25, 2014, 13:31:24 by brushedmoss »
« Reply #5 on: March 25, 2014, 13:45:15 »
watercooled *
Posts: 11

I'm considering upgrading to an APU1C myself but first I'd like to know what temperatures and power consumption are like - I've seen a few AC power measurements with a fully kitted-out system but I'll likely be using it with just an SD card and no wireless adapter.

Do you have a way of measuring the power consumption yourself? Or the temperature for that matter - e.g. does the core itself and/or the aluminium case get hot (I'm wondering what thermal transfer through the heatspreader and thermal pads is like).

I made some measurements myself on the ALIX 2D13 for comparison, if anyone's interested: https://forum.pfsense.org/index.php?topic=59555.msg386078#msg386078
« Reply #6 on: March 25, 2014, 13:56:44 »
brushedmoss ****
Posts: 446

there is no access to the temperature data from m0n0wall, and I don't have anything clever to measure temperature with than tell you if it seems warm or not Smiley , when testing a 800mbps copy for 10 mins, it didn't seem much above room temperature, warm but not hot. (yes it's subjective)

to measure current , I'd have to make a breakout wire to insert my meter, and run the tests again, I don't think i'll be doing that any time soon though, I can find out the rating of the adapter if thats any use to you (plus I have msata, wireless and 3g installed)
« Reply #7 on: March 25, 2014, 14:13:54 »
watercooled *
Posts: 11

Thanks. Smiley

It's comforting to know it's not getting hot; even my 2D13 gets lukewarm. I read some claims the CPU was getting to >80C on the beta board (hence my concerns about power consumption too), but I'm not sure how that was measured e.g. it could have been with no heatsink.

No worries if you don't have any power measurements to hand, I just asked on the off-chance. I know the ballpark numbers i.e. <10W idle, and the PC-Engines website claims ~6-12W depending on load.
« Reply #8 on: March 25, 2014, 14:17:15 »
brushedmoss ****
Posts: 446

i'll see if I can get a loan of an infra red heat measurement tool.  but for home use (130 Mbps cable modem) it's not getting hot at all
« Reply #9 on: March 25, 2014, 16:19:06 »
mikeboss *
Posts: 11

46° Celsius warmest spot at the bottom of the case while idling with Sophos UTM installed. this is a production board with an AMD G series T40E.

regards,
michael
« Reply #10 on: May 25, 2014, 19:09:54 »
Daniel Kulesz *
Posts: 3

I have a ALIX board with m0n0wall and recently bought the APU board where I currently installed Linux. The APU board consumes 6.4W in idle (with just the mSSD-module inserted, no wifi, 3G etc.) including the AC adapter. It gets much warmer on the bottom than the ALIX board (that's why it needs the heatspreaders) and the sensors reading are around 50-55°C during an operation which causes around 40-60% CPU load.

For me, the APU board is a step backward in terms of coolness but on the other side it offers much more performance and support for more peripherals (3G, MiniPCIe etc.) than the ALIX board. If you would compare it in terms of power efficiency, you should limit the bandwidth on each port to 100MBit and then measure the power consumption. But since the ALIX hardly uses above 4W under full load while the APU already uses over 6W in idle, I would argue that the APU is much less efficient in low load scenarios.
« Reply #11 on: June 19, 2014, 20:08:51 »
l8gravely *
Posts: 2

So what do people think about upgrading from an old WRAP board to an APU1c board?  My WRAP is starting to die under load without any useful logs or output to the console.  Sigh... so maybe it's time to upgrade to the next good thing.  I like the WRAP board, I barely think about it.  But I'm a little worried about the APU1C's heat.  But I do like the performance options.

And I like that I could (sorry for the heresy!) think about trying out smoothwall or ipfire or some other  linux/bsd based firewall. 

Thoughts?
John
« Reply #12 on: June 20, 2014, 15:32:13 »
watercooled *
Posts: 11

I got my APU a few weeks back and I'm happy with it overall. Like some other people have done, I decided to flip it over so the 'hot' side is facing upwards to allow better air circulation. This doesn't seem to make a huge difference, maybe 5C or so, but it doesn't hurt IMO. My IR thermometer currently measures 43C at the hottest spot on the bottom, ~21C ambient temperature. I really don't think the temperature is such a big deal; yes the Geode was very cool running, but many off-the-shelf routers get far hotter and use more power than the APU at the end of the day.

Power consumption and temps do seem slightly lower under Linux vs BSD, roughly a watt or so at idle based on my measurements i.e. 5 vs 6 Watts DC with just an SD card attached, no LAN cables or anything. However the power measurements fluctuate depending on CPU temperature so take the exact numbers with a pinch of salt.

Personally m0n0wall remains my preferred distro, not least because the traffic shaper is excellent. On that subject, I found some of Linux Distros basically just have an on/off switch for QoS with very little configurability built into the web interface. The one on pfSense has quite a learning curve, and while I need to do some more testing to figure it out, I found it would needlessly drop a lot of packets when the connection was heavily loaded (I suspect it has something to do with ECN) and not keep latency under control nearly as well as m0n0. Also m0n0 allows equal bandwidth sharing by IP with a checkbox, you'd need to add a load of custom rules to pf to achieve the same thing.

m0n0wall also seems to achieve higher throughput on given hardware vs pfSense, but neither would be  bottleneck for my current connection on the APU.
« Reply #13 on: June 21, 2014, 15:24:31 »
l8gravely *
Posts: 2

watercooled,

Thanks for the input.  I ended up ordering an APU yesterday, and with this change I'll take the opportunity to also move to m0n0wall 1.8 which my wrap wouldn't support.  I actually tried at one point running dd-wrt on a Netgear WND3700v2, but I hated their mDNS setup for a local internal DNS.  Adding new hosts was a total pain, where m0n0wall made it simple to add stuff.

Mostly  I just want a good firewall with internal services on demand, so I'm good with m0n0wall.

John
 
Pages: [1]
 
 
Powered by SMF 1.1.20 | SMF © 2013, Simple Machines