Operating temperature


#42

Thanks @backon - great solution!


#43

Well, generally, why to do it, there is no reason for that. Only may be in case of overclocking.


#44

operating time of the HW parts? I am not sure, why its so hot in idle (TP Link with OpenWRT basics doesnt heat any way, and more heat=more power consumption), but I will do the same just for better feeling for the parts :slight_smile:


#45

which TP-link has processor from Omnia?


#46

Thats not so bright question, you know :slight_smile: But I would say, why should more powerfull processor necessary heating more, when idling? I mean so “much” more…


#47

Of course, these things are just lets say better feeling.
Anyway, higher temperature i guess s logic. Because its just different processor. its normal. Important is that it fits to vendor specs. Thats all. But overclock, that would be another story of course. I know on Cisco routers i had this possibility of overclocking, here on omnia i am not sure. And about consumption - cooling does NOT reduce consumption.


#48

Yes, I can agree with different specs… But the heat affects also parts around… And I am person, which usually buy stuff, take really care, and things are with me a long time :slight_smile: Considering my Omnia will most of time idling, I would like to have at least heat taken out from PCB, if I will fail with lowering idle temperature through keeping software core basic.

And consumption yes, but when you running a similiar config (understand, TP Links are 10/100), noone is home (router is really IDLE), then I would expect less heat = less power consumption on idle mode (considering far less and older HW could take runtime within 0.07 CPU 24/7 average usage in my case - TP Link 741ND) .

Maybe I will sound silly, but you know what I am trying to say… Old HW, similiar software, and it performs without trip, with finger in nose and low heat/power consumption, I wouldnt expect far more power/heat increase with better HW, in idle state.

I say nothing about stress or something, then TP Link will goes down comparing, of course.


#49

You are just comparing uncomparable. That’s all. But this is no issue, its OK.

Temperature in idle is NOT solved by cooling, but by good drivers. If processor in idle should have less temperature (and again, that’s question of vendor, not you or me), so IF idle temperature should be lower, then there is a big space for drivers / firmware optimalisation, not for cooling. I don’t know, how many electronic devices you had in hands, but belive me, in electronic devices are really lot hot components, running years and you even don’t know about that. Only difference here is, that you have thermal sensor on processor, so you know the temperature. If you would not know that, you well “sleep like baby” even with current idle temperatures for years, without any cooling :slight_smile:

Anyway generally, of course any better cooling is generating less temperature on thermal sensor, then without cooling. That’s not the point.


#50

so IF idle temperature should be lower, then there is a big space for drivers / firmware optimalisation

agree. And still I am curious, why should be power consumption (if we would guess from heat) so much more in idling. You said its the specs. I reply: Hmm, OK, but, why so lot during idle? And I know, thats not fault of Turris creators or something (if its not the case bad swe/hw design which they can affect), but chip vendor… Just my experience with CPU/GPU chipset are: no load, very nice temp…

If you would not know that, you well “sleep like baby”

I noticed here reports, that case become significantly warm… So maybe not :slight_smile:

Anyway, I am not capable to optimise drivers anyhow, but I can do (according this thread) better job in heat “drain off”. So its not sensenless for me, as you suggest. That was my point… And after all, good feeling is sometimes priceless :slight_smile:


#51

You dont understand me at all.

  • I never said its senseless. I just said, that if temperature is in ranges of vendor’s spec, nothing can happend and its only psychological. Of course 40C looks better then 60C, but if has no impact, then no relevant story. If you will lower from 60 to 40, then its better then from 60 to 50. but worse then from 60 to 10

  • My box is not warmer as any other device like my mobile phone during light work, set-top-box, when switched on, TV, when switched on, etc. And I trust vendor of TV, that temperatures inside device are OK.

  • CPu/GPu processors - that’s completely different story. Its about drivers, not cooling again. I don’t know the omnias chipset specs in detail, but what I know, power consumption and power-status levels are really sophisticated in GPU / CPU. Because its oriented on that. I never heard something about this kind of focus on routers. I guess, chipset in your TV probably might have also lower temperatures, but I am nt sure, if you checked in detail your TV and mounted some cooling, I don’t think so.

Anyway, what is interesting for me, is chipset support. If omnia chipset has support for more sophisticated operations with power consumption - performance - generated heat, would be great to be implemented as nice option. In CPU/GPU world, chips are underclocked to reduce performance/consumption for example. I am not sure, if chipset in omnia supports that, or not. This is I thing more interesting thing. At least for me. From my experience, I had such routers from CISCO, you were able underclock / overclock. That would be nice. Much more relevant, then cooling.

But its really point of view, of course best is to have just omnia desk, out of box and big active cooler next to it. Then temperature will be even lower, then in your box.


#52

I didnt know that. If yes, that would be a great customize feature… But again, I agree and more dont expect such. It would be much picky for user :slight_smile:

If you are right about CPU/GPU vs Router CPU, then TP Link must be really “underclocked”, but still route with a big power reserve in my case (just little home network, few PCs, 2-3 Rpis…)


#53

It was possible on CISCO 3000 with DD-WRT. I am sure, that was supported also for other chips and routers. I think, routers does not have something like “idle” status. I guess they are running all the time on max frequency, its not like on PC or mobile device. And that’s the main difference. But on the other hand, omnia is may be more Linux-based pc then standard router, so might be there is chipset support. Good question for omnia administrators, may be they will “cook” sime nice features in this, who knows.


#54

I really don’t believe your TP-Link router works on same frequencies like omnia chipset. That counts, nothing else. You have to compare frequencies.


#55

Well, I think it is also purpose what counts… I could be minority with using Omnia as “just router”, I am aware of it, and no complains. Its beneficial to me, see that propably omnia is running by default (and consuming more power+heat more) at much much higher level than I need. Maybe I will consider not to strip software feautures so much, if it doesnt make power consumption different. Will see. I really though that it will be lowered just by less stress.

My TP-Link is most propably atheros 400MHz, and I am aware of difference between him and Omnia capabilities. Just commented his purpose: routing, and it does very, very well (openwrt, few additional packages, no ipv6).


#56

Well and now I am just guessing, but your TP-LINK is running at 400 MHz all the time. Does not matter, if idle, or under stress. But because its only 400 MHz, it cannot generate same heat as omnia chip. That’s it. And if your TP-LINK would underclock itself, then probably basic routing work time will be ±same, with even lower heating / consumption , then your TP-LINK has now. Temperature of your TP-LINK is lower, but does not mean, that cannot be even more lower. Its never-ending cycle. You never know, what temperature is that correct one. That’s why we have vendor specs.


#57

Even smartphone SoC go over 100 degrees celsius. They will step down when the outer case reaches around 40 degrees celsius. Their own temperature is kept at their specified limit. This is how it works. Did you ever think about a big cooler for your smartphone? Tablet? Notebook? Some smartphones take the other way and underclock their chips to get longer battery runtime and less step down actions.


#58

Really there are no reasons, apart from psychological, if the board had proper thermal design, but I will sleep better, if that chip runs juuuust a little bit colder.


#59

well, mine was running pretty much at 86°C all the time … after raplacing the the whitish goo, the passive was sticked with to the SoC for aforementioned arctic pad, I’m now on around 70. …very happy with the result.


#60

Mine sits at around 70C under usual load, sitting in a rack that’s at around 10C as measured by the ambient sensors I have there. Perhaps time to add an external fan to blow air through the case holes. I don’t care about noise since it all sits in the basement and the server fans aren’t exactly quiet.


#61

Backon, how do you separe the block from the chip? Mine is sticked as glue, and I dont want to break anything by hard force…