Streacom's FC10 and Nano150: Building a Fanless Ivy Bridge HTPC
by Ganesh T S on December 22, 2012 3:30 AM EST- Posted in
- HTPC
- Fanless
- Ivy Bridge
Thermal Performance
Evaluation of the thermal performance of passive systems is absolutely essential because it is quite common for improperly designed thermal solutions to not prevent processors from reaching their maximum permissible junction temperature. Once this temperature is reached, thermal throttling is activated. The processor first reacts by lowering its operating frequency. If the cooling solution is bad, it might even result in the processor shutting itself down.
The Streacom OEM solution used in the Aleutia Relia wasn't very effective (though Streacom refused to comment on those results). However, their previous chassis / cooling solutions had always been effective in operation, and we weren't unduly worried about the FC10 with respect to this aspect. That said, we did carry out the same experiments performed with the Aleutia Relia, but restricted ourselves to room temperature (72 F = 22 C) tests only.
Our loading test consists of running Prime 95 in maximum power consumption mode along with Furmark in Stability Test mode. Instead of running the test for 20 minutes, we subjected the unit to 12 hours of stress continuously.
Power consumed at the wall was recorded for the first 90 minutes or so (and we found that it settled down to around 72 W beyond that). Unlike the Relia where thermal throttling was activated and resulted in power consumption at the wall going down after some time, we find that there is no thermal throttling at play in this system. The load and CPU frequencies were presented in the Relia review, but we won't present them here since the load stayed at 100% and the frequency of the cores was always at 3.3 GHz throughout the course of the stress test.
The CPU core temperatures are presented in the graphs above. Interestingly, the second core was always a couple of degrees cooler than the first core. The maximum junction temperature of the Core i3-3225 is 105 C, and the FC10 cooling mechanism was able to keep it at a safe temperature.
Another interesting aspect is how fast the unit is able to get back to the idle temperature after removal of the processor load. The graph below shows that the unit gets back to the idling temperature within 90 minutes.
We conclude this section with a thermal profile of the case and the heat sink when the unit is being subjected to full load using Prime 95 and Furmark. All the temperatures are in Farenheit.
The temperature on the lower CPU mount noted above was actually on top of the FC10 upper CPU mount (Intel-type). All these values were recorded with the unit being kept in a still room with plenty of open space around it.
Facebook
Twitter
RSS
63 Comments
View All Comments
colonelciller - Sunday, December 23, 2012 - link
Does the i3-3225 support 23.976 playback?Aikouka - Sunday, December 23, 2012 - link
Last I recall, Intel's drivers do have a "24 FPS" setting, but it is not true 23.976. However, my information may be out of date!casteve - Saturday, December 22, 2012 - link
Thanks for the review, but what are the units of time in the graphs? seconds, minutes, mayan long count?ganeshts - Saturday, December 22, 2012 - link
Apologize for the oversight. The time unit is in (s). The first graph is not very clear (I had actually let our power logger script run for 100 minutes = 6000s). For the loading temperature graphs, I had indicated in the text that the burn-in testing was run for 12 hours (=43200s).kyuu - Saturday, December 22, 2012 - link
First paragraph on page 3:"There was a toss up between building a Trinity-based testbed and a Ivy Bridge-based testbed. In the end, the fact that Trinity emerged as being a capable madVR candidate (with software based decoding), and the fact that madVR recently introduced DXVA scaling (an upside for Intel since its offerings weren't fully madVR capable earlier, and something that we wanted to test out) persuaded us to go for an Intel-based testbed."
Am I missing something or is this messed up? You describe (what I assume are) positive aspects of Trinity, then use that as the justification for using Ivy Bridge? Huh?
ElvenLemming - Saturday, December 22, 2012 - link
The way I interpreted that was that because the madVR support is newer and less-tested for Intel, they decided to go with an Intel build here so they could use the same machine for future madVR testing.The paragraph wasn't worded very clearly, but that's what I got out of it.
ganeshts - Saturday, December 22, 2012 - link
Yes, I could have conveyed that in a better manner, but ElvenLemming's interpretation is right. In our Trinity and Ivy Bridge reviews, we found that 720p60 H.264 streams gave trouble to HD 4000. Recently, madVR introducied DXVA scaling, and I have it from very reliable sources that Mathias (madVR developer) is working closely with Intel. So, we wanted to have the right platform to evaluate these developments.kyuu - Sunday, December 23, 2012 - link
Ah got it. Thanks for the clarification.Schafdog - Saturday, December 22, 2012 - link
What's the idle power of the system?ganeshts - Saturday, December 22, 2012 - link
That is coming in Part 2 of the review :) As a sneak peek, I can say it is around 28 W. Power measurement during Blu-ray playback and other workloads will be dealt with in detail in the next series installment.