[Invalid] - OrangePi4 - RK3399 - Ubuntu 20.04 - High frequencies, overheating and lack of power.

[p0rc0_r0ss0]

I've built custom Linux 5 kernel for Oprange PI 4 including only options required for this board and module drivers for peripheral devices. 

Long story shor: It works, it boots, it runs, it's hot.

 

Now I _think_ I've located cause of overheat:

Linux 5.*.*
08:58:34: 1800/1416MHz  2.84  50%   1%   8%   0%  40%   0% 73.3°C  0/5
08:58:40: 1800/1416MHz  2.85  28%   0%   7%   0%  20%   0% 73.3°C  0/5
08:58:45: 1800/1416MHz  3.35  62%   0%   8%   0%  53%   0% 75.6°C  0/5
08:58:50: 1800/1416MHz  3.96  48%   1%   9%   0%  37%   0% 73.9°C  0/5
08:58:55: 1800/1416MHz  4.04  42%   0%   8%   0%  33%   0% 76.9°C  0/5
08:59:00: 1800/1416MHz  4.20  71%   1%   8%   0%  61%   0% 75.6°C  0/5

Compared to this, working on top capacity ~54 degrees:

Linux 4.*.*
10:05:44: 1200/1200MHz  2.44  31%   1%  19%   0%  10%   0%  0/5
10:05:49: 1200/1200MHz  2.73  28%   2%  18%   0%   7%   0%  0/5
10:05:55: 1200/1200MHz  2.67  32%   6%  23%   0%   1%   0%  0/5
10:06:00: 1200/1200MHz  2.61  31%   6%  22%   0%   1%   0%  0/5
10:06:05: 1200/1200MHz  2.56  32%   6%  23%   0%   2%   0%  0/5
10:06:10: 1008/1200MHz  2.44  23%   1%  19%   0%   2%   0%  0/5
10:06:15: 1008/1200MHz  2.80  27%   0%  17%   0%   9%   0%  0/5

At the same time on 5.* kernel my USB HDD starts to make strange noises of re initialization.

 

As I get it, 4.* stock kernel is limited by max frequency, which results less heat and less power consumption (obviously). When I run it full speed my USB voltage diminishes.

 

Are any heads up on system tuning for this board?

[Werner]

Your issue report is not a valid bug report per the Armbian bug reporting instructions (https://www.armbian.com/bugs).  With limited resources the Armbian project is only able to spend time on issues where all the requested information has been provided and for only the boards/images/software that are supported.  Your report is invalid for one or more of the following reasons (non-exhaustive list):

 

• it is for an unsupported board or image (CSC/EOS/WIP/edge)
• it is for software that is not supported (such as userspace modules installed on top of the core operating system)
• it has been logged in the wrong forum (for example requests for help that are not actual bug reports)
• it lacks requested data (armbianmonitor output)
• it could have been easily solved by a quick search and/or reading documentation

 

Please review what you have submitted and the bug logging instructions (https://www.armbian.com/bugs) and either add the required information or open a new topic in the correct forum (such as Common issues / peer to peer technical support or General chit chat)

[p0rc0_r0ss0]

Thank you for your reply,

My post was not intended as bug report, initially I placed it in "Reviews, Tutorials, Hardware hacks" thread.

The idea was to share hardware tuning settings, best practices and kernel tweaks.

 

This post was made not to disturb core Armbian team members, but rather accumulate live user experience. I think everybody will benefit with this approach.

 

It would be not ideal to search for another forum and share opinions on how to tweak Armbian on ARM hardware. In fact, it may be productive to have separate forum section to share usage experience. It might lower volume of basic questions addressed to core team members.

Thank you.

[NicoD]

@p0rc0_r0ss0You didn't share much information. As Werner mentioned. The output of armbianmonitor would tell a lot. You didn't mention if or what cooling you use.
I'm a bit puzzled by the information you did share. At kernel 5.x you show a CPU frequency of 1800/1416MHz. That's the stock clockspeeds for RK3399.
At kernel 4.x it's 1200/1200MHz and 1008/1200MHz.
That looks like it's is throttling heavily. No temperatures are shown with it.

Do you have any heatsink on it? The RK3399 does get very hot when not cooled well.

17 hours ago, p0rc0_r0ss0 said:

When I run it full speed my USB voltage diminishes.

That gives me the hint that your power supply might not be sufficient. An RK3399 maxed + peripherals can consume +3A.
I would suggest you look for a big heatsink, a better power supply. And then use sbc-bench to see how the system behaves.
If you don't need the maximum horsepower you can underclock the system. It will consume a lot less, and run a lot cooler.
You can find sbc-bench and set cpu frequency with armbian-config.

[p0rc0_r0ss0]

Benchmarks were made on current hardware as a starting point.

System A sbc-bench output: http://ix.io/3qAB

System B: http://ix.io/3qAU

Right now I have pretty small passive heatsink on board.

If I get it correctly, new kernel is more effective, but in both cases "fast" A72 cores got throttled down.

Yeah, I was thinking of getting 5Amp power supply, right now it's 4A, but I have doubts if it's legit 4A.

[NicoD]

That is very heavy throttling.

System A

System health while running 7-zip multi core benchmark:

Time       big.LITTLE   load %cpu %sys %usr %nice %io %irq   Temp
23:32:04: 1800/1416MHz  4.91  18%   0%  14%   0%   2%   0%  68.3°C
23:32:27: 1416/1416MHz  4.62  80%   1%  78%   0%   0%   0%  71.7°C
23:32:49: 1200/1416MHz  5.63  89%   1%  88%   0%   0%   0%  72.8°C
23:33:10: 1200/1416MHz  5.44  88%   2%  85%   0%   0%   0%  72.8°C
23:33:30: 1416/1416MHz  6.00  93%   1%  92%   0%   0%   0%  73.3°C
23:33:50: 1200/1416MHz  5.78  82%   1%  80%   0%   0%   0%  73.3°C
23:34:11: 1200/1416MHz  6.14  91%   1%  90%   0%   0%   0%  73.9°C
23:34:32: 1200/1416MHz  6.09  84%   1%  82%   0%   0%   0%  74.4°C
23:34:52: 1416/1416MHz  6.61  99%   1%  97%   0%   0%   0%  74.4°C
23:35:12: 1200/1416MHz  6.24  80%   1%  79%   0%   0%   0%  75.6°C
23:35:33: 1200/1416MHz  6.16  90%   1%  88%   0%   0%   0%  75.6°C
23:35:53: 1200/1416MHz  6.04  82%   1%  80%   0%   0%   0%  75.6°C
23:36:14: 1200/1416MHz  6.60  98%   1%  96%   0%   0%   0%  75.6°C

It seems to start throttling at 70C. That seems strange to me. But that's is not a big problem. Big core goes down to 1200Mhz.
 

System B
 

System health while running 7-zip multi core benchmark:

Time       big.LITTLE   load %cpu %sys %usr %nice %io %irq   Temp
00:42:38: 1608/1416MHz  5.33  14%   0%  13%   0%   0%   0%  84.4°C
00:43:01:  600/1416MHz  5.46  93%   0%  92%   0%   0%   0%  84.4°C
00:43:26:  816/1416MHz  5.23  70%   0%  70%   0%   0%   0%  85.0°C
00:43:47:  816/1416MHz  5.03  78%   1%  77%   0%   0%   0%  84.4°C
00:44:07: 1008/1416MHz  5.07  76%   0%  76%   0%   0%   0%  84.4°C
00:44:28:  816/1416MHz  5.08  87%   2%  85%   0%   0%   0%  84.4°C
00:44:48:  600/1416MHz  5.47  91%   1%  90%   0%   0%   0%  85.0°C
00:45:14:  600/1416MHz  5.32  69%   1%  67%   0%   0%   0%  85.0°C
00:45:39:  816/1416MHz  5.29  78%   0%  77%   0%   0%   0%  84.4°C
00:46:01:  816/1416MHz  5.03  75%   0%  74%   0%   0%   0%  84.4°C
00:46:21:  816/1416MHz  5.43  92%   1%  91%   0%   0%   0%  85.0°C
00:46:42: 1008/1416MHz  5.74  71%   1%  69%   0%   0%   0%  85.0°C
00:47:05:  600/1416MHz  5.93  92%   1%  91%   0%   0%   0%  85.0°C
00:47:29:  600/1416MHz  5.40  70%   1%  69%   0%   0%   0%  85.0°C
00:47:51:  816/1416MHz  5.02  75%   0%  74%   0%   0%   0%  84.4°C
00:48:12: 1008/1416MHz  4.98  80%   0%  79%   0%   0%   0%  84.4°C
00:48:37:  600/1416MHz  5.54  87%   1%  85%   0%   0%   0%  84.4°C
00:48:58: 1008/1416MHz  5.56  72%   1%  70%   0%   0%   0%  84.4°C
00:49:23:  600/1416MHz  5.52  94%   2%  91%   0%   0%   0%  85.0°C

It goes up to 85C and throttles very heavy down to 600Mhz.

So something is seriously wrong with your System B. Even in single core 7-zip it is throttling at 85C. That is not normal.
System A throttles at a lower temperature. But it doesn't seem it is running that much hotter than expected with a small heatsink.
But compared to that, B throttles lower while being at a lot higher temperature.

What images are you using, what did you change?

 

Another tip would be to check your voltages at the source, and at the USB ports with a USB voltage meter.
And again, the output of armbianmonitor.

 

Not enough information is given to know what is happening. But it seems to me that system B should not be used. Something is seriously wrong with it.

PS.
This is how it should look from a healty RK3399 with ok cooling. http://ix.io/3qt9

[p0rc0_r0ss0]

First of all, thank you for your help. Much appreciated.

 

Test system description

Hardware is OrangePi4 4Gb + external USB 2.5 HDD drive.

System A is stock OrangePi4 image. This one here.

System B is same hardware with Armbian built on Ubuntu 20.04 with custom kernel config.

It was pretty much first build of this kernel and I guess kernel needs a bit tuning.

Now I'm rebuilding it with

Quote

CONFIG_ARM_PSCI_CPUIDLE=y

Which seems to cause system ignore idle states.

 

Forged heat sink from old PIII CPU cooler, got this results. No more throttling on first image. http://ix.io/3qDo

 

The problem with first image for me is that it doesn't support modules like netfilter and overlay2. It also updates from orangePi repo which sometimes is very slow.

[NicoD]

9 hours ago, p0rc0_r0ss0 said:

Forged heat sink from old PIII CPU cooler, got this results. No more throttling on first image. http://ix.io/3qDo

That looks a lot better.

Could you try a default armbian image and see how temperatures are with it? https://www.armbian.com/orange-pi-4/

I do not own the OPi4. I'm not aware of the state of it. But those temperatures I've seen even badly cooled seemed very weird. I'd like to know if it is running high with a default image. 

[p0rc0_r0ss0]

I did two tests:

1. I made my own custom build with "CONFIG_ARM_PSCI_CPUIDLE=y" added and got this results: http://ix.io/3qEx

2. Then made image with prebuilt kernel from repo. http://ix.io/3qJo

This compared I get idea that my kernel config was not that bad after all.

I checked power supply and it is 5V 3A, I have a feeling that on v5 kernel my board lacks power. Made this conclusion because on new kernels (petty much all v5) USB disk makes this ticking sound, probably lacks power.

 

I want to try install active cooling.

[p0rc0_r0ss0]

Ok, fixed cooling, added 10A power supply. I guess cause of system instability somewhere in kernel config.

I'm using stock one.

Any ideas how to resolve this issue?

[thc013]

a 3A power supply is enough

 

you can check L2101 inductor the opi 4 is little bad with noisy power

 

and active cooling is needed for working under heavy load on a rk3399

 

i compile gentoo on a opi4 wit all usb populated and a mini pcie wifi

(yeah rust,gcc,clang spidermpnkey is no fun )

idle 30c /full load 60 c

 

5.10 is the most stable