Jump to content

Random issues with nightly images on Orange Pi PC2


Bart

Recommended Posts

Hello

 

I installed today image version: Armbian_5.24.170129 (CLI version)

Image starting fine but after some time i'm unable to login via ssh and hdmi showing something like this:

gRxgqjV.jpg

 

Any ideas?

 

Its on sandisk card class 10.. (brand new )

but kingston (used for 1,5 y)card also class 10 warking fine,

Edited by tkaiser
Splitted from development/testing thread!
Link to comment
Share on other sites

Any ideas?

 

Its on sandisk card class 10.. (brand new )

but kingston (used for 1,5 y)card also class 10 warking fine,

 

Well, I vote for just single SD card issue + may be your PSU is not powerful enough/noisy. I'm using old class 4 card for test, no issues so far.

Link to comment
Share on other sites

Well, I vote for just single SD card issue + may be your PSU is not powerful enough/noisy. I'm using old class 4 card for test, no issues so far.

 

Hym its 3Ampers so i think that is enought.. i have one more problem i'm using pc2  for seedbox (downloading torrents via rtorrent)

1KXKEOs.png

 

Its rly good performance but after some time pi is rebooting... temp are good i think because there is no more than 50 Celsius..

Link to comment
Share on other sites

Maybe psu isnt really capable of 3A..

Hymm it can be true...

But i'm dowloading to external HDD (which have own power supply 12v), but it can take also some power from usb...

Can you recmomend good power supply ?

Link to comment
Share on other sites

It's unbelievable. 3 days ago I wrote in the thread you started to pollute with support questions: 'This is neither an announcement of Armbian support for this device nor anything a typical end user should play with'. So please do not disrupt development/testing efforts in the main thread and stay with your support talk here.

Link to comment
Share on other sites

Hymm it can be true...

But i'm dowloading to external HDD (which have own power supply 12v), but it can take also some power from usb...

Can you recmomend good power supply ?

 

MeanWell is great, those guys design PSU's for "other guys", including AAA-brands.  Also this one mentioned by @tkaiser is the "official one", so the final choice is up to you.

Link to comment
Share on other sites

 

It's unbelievable. 3 days ago I wrote in the thread you started to pollute with support questions: 'This is neither an announcement of Armbian support for this device nor anything a typical end user should play with'. So please do not disrupt development/testing efforts in the main thread and stay with your support talk here.

 

Maybe rename this thread to something like "Unofficial H5 support" so noobs won't trash dev thread.

Link to comment
Share on other sites

Maybe rename this thread to something like "Unofficial H5 support" so noobs won't trash dev thread.

 

There is no such thing like 'H5 support' currently and we're talking here just about another occurence of either SD card or power supply issues (or both -- pretty boring especially from the perspective of those guys having written the 'Getting Started' recommendations and wasted their time to implement easy logfile uploading -- see signature below).

 

if people would read/search more and write less this forum would be a lot more useful than now :(

Link to comment
Share on other sites

Can someone explain me how cooling table (or whatever kernel uses) works when determining frequencies ? Because lookig at the frequecies, load and temps, i can't see how frequencies are determined.

 

FOr example... CPU temps are below  50C, there is ton of load on the cpu, yet cpu somehow throttles down to below 1Ghz for some reason, even though there is plenty of thermal headrom available. Is this intentional, so that cpu stays in more efficient operation range (less heat for a given workload) or is something else at play ?

Link to comment
Share on other sites

FOr example... CPU temps are below  50C, there is ton of load on the cpu, yet cpu somehow throttles down to below 1Ghz for some reason, even though there is plenty of thermal headrom available. Is this intentional, so that cpu stays in more efficient operation range (less heat for a given workload) or is something else at play ?

It's possible that there are issues with the current DVFS/THS code. But output of "armbianmonitor -m" may be helpful to see the full picture.

Link to comment
Share on other sites

Are you referring to CPU utilization or average load when you're talking about 'load'? Anyway without 'armbianmonitor -m' output or RPi-Monitor graphs it's pretty useless...

 

 

I will do a test later today to better illustrate this...

Link to comment
Share on other sites

So here are my results. I was mostly doing web browsing via firefox. PC2 is enclosed in offical case with top removed and small chipset heatsink installed. CPU governer is schedutil. It's really odd, because sometimes when scrolling, there is lots of load on the cpu, yet governet does not respond well. To my understanding, THS/governer should allow boosting to 1.3Ghz a lot more often given how i was always staying below 75C

Time        CPU    load %cpu %sys %usr %nice %io %irq   CPU
20:17:00: 1296MHz  1.50  25%   6%  11%   0%   7%   0% 38.3°C
20:17:05:  480MHz  1.38  22%   5%   9%   0%   6%   0% 40.5°C
20:17:10:  480MHz  1.27  22%   5%  10%   0%   5%   0% 38.6°C
20:17:16:  648MHz  1.25  22%   5%  11%   0%   5%   0% 38.0°C
20:17:21: 1056MHz  1.23  23%   5%  12%   0%   5%   0% 40.9°C
20:17:26:  480MHz  1.21  24%   4%  13%   0%   5%   0% 42.1°C
20:17:31:  480MHz  1.19  24%   4%  14%   0%   4%   0% 37.6°C
20:17:36:  480MHz  1.10  23%   4%  13%   0%   4%   0% 36.3°C
20:17:41:  480MHz  1.01  22%   4%  13%   0%   4%   0% 37.0°C
20:17:46:  480MHz  0.93  21%   4%  12%   0%   3%   0% 40.0°C
20:17:51:  480MHz  0.85  20%   3%  12%   0%   3%   0% 39.9°C
20:17:56:  648MHz  0.79  20%   3%  12%   0%   3%   0% 39.5°C
20:18:01:  648MHz  1.04  21%   3%  14%   0%   3%   0% 38.3°C
20:18:07:  480MHz  1.12  22%   3%  15%   0%   3%   0% 39.9°C
20:18:12:  648MHz  1.11  23%   3%  15%   0%   3%   0% 39.6°C
20:18:17:  960MHz  1.18  24%   3%  16%   0%   3%   0% 42.9°C
20:18:22:  480MHz  1.09  23%   3%  16%   0%   2%   0% 41.6°C
20:18:27:  648MHz  1.00  23%   3%  16%   0%   2%   0% 38.6°C
20:18:32:  960MHz  1.08  24%   3%  17%   0%   2%   0% 43.4°C
20:18:37:  648MHz  1.15  25%   3%  18%   0%   2%   0% 40.3°C
20:18:42:  960MHz  1.14  25%   3%  19%   0%   2%   0% 41.1°C
20:18:47:  480MHz  1.29  25%   3%  19%   0%   2%   0% 38.3°C
20:18:52:  480MHz  1.35  25%   3%  18%   0%   2%   0% 37.3°C
20:18:58:  480MHz  1.24  24%   3%  18%   0%   2%   0% 41.8°C
20:19:03:  480MHz  1.14  24%   3%  18%   0%   2%   0% 40.2°C
20:19:08:  480MHz  1.05  23%   3%  17%   0%   2%   0% 38.7°C
20:19:13:  480MHz  0.96  23%   3%  17%   0%   2%   0% 37.9°C
20:19:18:  480MHz  0.97  22%   3%  17%   0%   1%   0% 38.3°C
20:19:23:  480MHz  0.89  22%   3%  17%   0%   1%   0% 38.6°C
20:19:28: 1296MHz  0.82  22%   3%  16%   0%   1%   0% 42.8°C
20:19:33: 1296MHz  0.75  21%   3%  16%   0%   1%   0% 38.9°C
20:19:38:  480MHz  0.69  21%   3%  16%   0%   1%   0% 37.6°C
20:19:43:  480MHz  0.64  21%   3%  16%   0%   1%   0% 39.3°C
20:19:49: 1296MHz  0.59  20%   3%  15%   0%   1%   0% 39.0°C
20:19:54:  480MHz  0.54  20%   3%  15%   0%   1%   0% 43.2°C
20:19:59:  480MHz  0.66  20%   2%  15%   0%   1%   0% 39.0°C
20:20:04:  480MHz  0.60  19%   2%  15%   0%   1%   0% 39.0°C
20:20:09:  480MHz  0.56  19%   2%  14%   0%   1%   0% 39.2°C
20:20:14:  768MHz  0.75  19%   2%  15%   0%   1%   0% 42.7°C
20:20:19: 1152MHz  0.77  20%   2%  15%   0%   1%   0% 43.4°C
20:20:24: 1296MHz  0.79  20%   2%  16%   0%   1%   0% 51.2°C
20:20:29: 1296MHz  0.73  21%   2%  16%   0%   1%   0% 45.0°C
20:20:34: 1296MHz  0.75  21%   2%  17%   0%   1%   0% 51.4°C
20:20:39: 1296MHz  0.69  21%   2%  17%   0%   1%   0% 45.8°C
20:20:44:  480MHz  0.63  21%   2%  17%   0%   1%   0% 37.7°C
20:20:49:  648MHz  0.74  21%   2%  17%   0%   1%   0% 42.9°C
20:20:55: 1008MHz  0.76  22%   2%  17%   0%   1%   0% 41.3°C
20:21:00: 1296MHz  1.10  22%   2%  18%   0%   1%   0% 53.5°C
20:21:05: 1296MHz  1.42  23%   2%  19%   0%   1%   0% 57.3°C
20:21:10: 1296MHz  1.46  24%   2%  20%   0%   1%   0% 53.5°C
20:21:15: 1296MHz  1.43  25%   2%  20%   0%   1%   0% 56.9°C
20:21:20: 1200MHz  1.63  25%   2%  21%   0%   1%   0% 55.7°C
20:21:25: 1296MHz  1.74  26%   2%  22%   0%   1%   0% 58.6°C
20:21:30: 1296MHz  1.84  26%   2%  22%   0%   1%   0% 60.1°C
20:21:35: 1296MHz  1.86  27%   2%  23%   0%   1%   0% 59.4°C
20:21:40:  648MHz  1.95  28%   2%  23%   0%   1%   0% 47.0°C
20:21:45:  480MHz  1.79  28%   2%  24%   0%   1%   0% 43.5°C
20:21:51:  816MHz  1.89  29%   2%  24%   0%   1%   0% 48.0°C
20:21:56:  480MHz  1.74  29%   2%  25%   0%   1%   0% 43.8°C
20:22:01:  768MHz  1.76  29%   2%  25%   0%   1%   0% 47.2°C
20:22:06:  912MHz  1.86  29%   2%  25%   0%   1%   0% 47.7°C
20:22:11:  480MHz  1.87  30%   2%  26%   0%   0%   0% 47.7°C
20:22:16:  792MHz  1.80  30%   2%  26%   0%   0%   0% 45.8°C
20:22:21: 1296MHz  1.82  30%   2%  26%   0%   0%   0% 49.0°C
20:22:26: 1296MHz  2.07  31%   2%  27%   0%   0%   0% 59.1°C
20:22:31: 1296MHz  2.15  31%   2%  27%   0%   0%   0% 55.9°C
20:22:37:  648MHz  2.13  31%   2%  27%   0%   0%   0% 46.9°C
20:22:42: 1008MHz  2.12  31%   2%  27%   0%   0%   0% 47.3°C
20:22:47: 1296MHz  1.95  31%   2%  27%   0%   0%   0% 54.8°C
20:22:52:  648MHz  2.04  32%   2%  28%   0%   0%   0% 46.4°C
20:22:57:  480MHz  1.87  32%   2%  27%   0%   0%   0% 45.8°C
20:23:02:  480MHz  1.88  31%   2%  27%   0%   0%   0% 44.1°C
20:23:07:  480MHz  1.81  31%   2%  27%   0%   0%   0% 44.8°C
20:23:12:  480MHz  1.67  31%   2%  27%   0%   0%   0% 43.8°C
20:23:17:  480MHz  1.69  31%   2%  27%   0%   0%   0% 44.0°C
20:23:22:  528MHz  1.64  30%   2%  26%   0%   0%   0% 45.4°C
20:23:28:  648MHz  1.51  30%   2%  26%   0%   0%   0% 46.3°C
20:23:33:  648MHz  1.71  31%   2%  27%   0%   0%   0% 44.5°C
20:23:38: 1008MHz  1.81  31%   2%  27%   0%   0%   0% 46.3°C
20:23:43:  912MHz  1.67  31%   2%  27%   0%   0%   0% 55.6°C
20:23:48:  864MHz  1.53  32%   3%  28%   0%   0%   0% 47.2°C
20:23:53:  480MHz  1.49  31%   2%  27%   0%   0%   0% 44.7°C
20:23:58: 1296MHz  1.55  31%   2%  27%   0%   0%   0% 47.6°C
20:24:03:  480MHz  1.43  31%   2%  27%   0%   0%   0% 42.9°C
20:24:08: 1296MHz  1.48  31%   2%  27%   0%   0%   0% 57.9°C
20:24:14:  480MHz  1.44  31%   2%  27%   0%   0%   0% 44.7°C
20:24:19:  648MHz  1.48  31%   2%  27%   0%   0%   0% 44.0°C
20:24:24:  960MHz  1.44  31%   2%  27%   0%   0%   0% 44.5°C
20:24:29: 1200MHz  1.41  31%   3%  27%   0%   0%   0% 48.2°C
20:24:35:  648MHz  1.62  31%   3%  27%   0%   1%   0% 44.4°C
20:24:40:  480MHz  1.57  31%   2%  27%   0%   1%   0% 43.8°C
20:24:46: 1296MHz  1.44  31%   3%  27%   0%   1%   0% 55.0°C
20:24:51:  648MHz  1.49  31%   3%  27%   0%   1%   0% 43.7°C
20:24:56:  528MHz  1.69  32%   3%  27%   0%   1%   0% 46.7°C
20:25:01:  648MHz  1.63  32%   3%  27%   0%   1%   0% 46.1°C
20:25:06:  912MHz  1.66  32%   3%  27%   0%   1%   0% 46.7°C
20:25:11:  480MHz  1.61  32%   3%  27%   0%   1%   0% 44.8°C
20:25:16:  480MHz  1.48  32%   3%  26%   0%   1%   0% 43.5°C
20:25:22:  528MHz  1.60  32%   3%  26%   0%   1%   0% 44.7°C
20:25:27:  480MHz  1.55  32%   3%  26%   0%   1%   0% 44.1°C
20:25:32:  480MHz  1.51  31%   3%  26%   0%   1%   0% 43.4°C
20:25:37:  480MHz  1.39  31%   3%  26%   0%   1%   0% 44.5°C
20:25:42:  912MHz  1.28  31%   3%  26%   0%   1%   0% 44.8°C
20:25:47:  480MHz  1.17  31%   3%  26%   0%   1%   0% 42.9°C
20:25:52:  480MHz  1.40  31%   3%  26%   0%   1%   0% 43.5°C
20:25:57:  648MHz  1.69  31%   3%  25%   0%   1%   0% 44.5°C
20:26:02: 1296MHz  1.55  30%   3%  25%   0%   1%   0% 47.6°C
20:26:07:  480MHz  1.43  30%   3%  25%   0%   1%   0% 43.1°C
20:26:13:  480MHz  1.31  30%   3%  25%   0%   1%   0% 43.5°C
20:26:18: 1296MHz  1.21  30%   3%  25%   0%   1%   0% 47.4°C^C

Link to comment
Share on other sites

Further looking at the "issue", it seems that schedutil must be a bit funky (at least for my use case). Or maybe low refresh rate of the armbianmonitor doesn't catch all frequency spikes from load.

 

In any case, setting to performance governor leaves the cpu at 1.3Ghz (as it should) but obviously at the expense of efficiency and heat.

 

Maybe a faster responding governer with temparature awarness could be handy in this case.

Link to comment
Share on other sites

One thing i must note, that  being a very weak SoC still, this thing is already quite usefull as a desktop "replacement". Firefox seems smooth (especially given how theres no accelaration whatsoever). Also it boots very very quickly (almost as fast as a desktop with ssd). Could be be attributed to fast samsung evo card i'm using, but legacy PC images needs much more time to boot with the same card.

Link to comment
Share on other sites

slow sd card (but evos arent actually that slow) shouldn't affect scrolling and scrolling take a lot of load. so in that instance, cpu clock should boost to 1.3Ghz.

 

in any case, setting to performance governor makes everything smoother, so you cant really blame anything other than cpu frequency scaling.

Link to comment
Share on other sites

slow sd card (but evos arent actually that slow) shouldn't affect scrolling and scrolling take a lot of load

 

Still you should try to get familiar with the stuff you're talking about (average load which is something different than CPU utilization -- at least on Linux!). So your 'use case' is scrolling. Please notice that Zador enabled xf86-video-fbturbo on OPi PC2 recently and in case you're able to measure scrolling performance comparing older nightly build with most recent would provide some useful feedback. 

 

Re scrolling please also keep in mind what the xf86-video-fbturbo author writes: https://irclog.whitequark.org/linux-sunxi/2016-03-19#15816125;

Link to comment
Share on other sites

Still you should try to get familiar with the stuff you're talking about (average load which is something different than CPU utilization -- at least on Linux!). So your 'use case' is scrolling. Please notice that Zador enabled xf86-video-fbturbo on OPi PC2 recently and in case you're able to measure scrolling performance comparing older nightly build with most recent would provide some useful feedback. 

 

Re scrolling please also keep in mind what the xf86-video-fbturbo author writes: https://irclog.whitequark.org/linux-sunxi/2016-03-19#15816125;

I'm just experimenting with it, I don't know what options are used to build the nightlies so we need to make sure that fbturbo is installed and is working fine. Also I'm not sure how much improvement can it provide (if any) on ARM64 and with framebuffer display driver.

Link to comment
Share on other sites

Still you should try to get familiar with the stuff you're talking about (average load which is something different than CPU utilization -- at least on Linux!). So your 'use case' is scrolling. Please notice that Zador enabled xf86-video-fbturbo on OPi PC2 recently and in case you're able to measure scrolling performance comparing older nightly build with most recent would provide some useful feedback. 

 

Re scrolling please also keep in mind what the xf86-video-fbturbo author writes: https://irclog.whitequark.org/linux-sunxi/2016-03-19#15816125;

 

Thank you for this, will give it a try.

Link to comment
Share on other sites

Tried the new build with fbturbo. Can't really say i see a noticable difference, if anything could be a placebo. Is there a way to quantify such performance ?

 

On a less related note; does allwinner h5 (or any other aw chipsets) support independed core clocking or is it a single frequency for all cores ?

Maybe if each core was individually clocked, this could help reducing heat.

Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
×
×
  • Create New...

Important Information

Terms of Use - Privacy Policy - Guidelines