Heatsink / fan vs. [ddrfreq]

[piknew]

Hi, if anybody can suggest if mentioned situation is OK? Thanks in advance.

 

Originally all my OPI boards have been running on 624 MHz for DDR frequency. 2 days ago I decided to remove fans from My Orange Pi+ 2e and Orange Pi+ 2 (following suggestions from forum as fan is maybe not required).

 

1. Orange Pi+ 2e (this is running all my services which is: nextcloud, mail server, svn and web server to access mentioned) - originally with heasink and fan (as you can see downtime on Oct 14th - when I have removed fan and keep only heatsink). But something happened - with fan ddr frequency was always 624 MHz. Without fan some time after reboot it is increasing frequency to 672MHz, for example:

 

[   10.850313] gmac0 gmac0: eth0: eth0: PHY ID 001cc915 at 0 IRQ poll (gmac0-0:00)
[   14.850329] PHY: gmac0-0:00 - Link is Up - 1000/Full
[   24.680720] ADDRCONF(NETDEV_UP): wlan0: link is not ready
[12328.080061] [ddrfreq] DDR: 624000KHz->672000KHz start
[12328.080074] [ddrfreq] current cpu is cpu2
[12328.081263] [ddrfreq] elapsed: 1041us
[12328.081276] [ddrfreq] DDR: 624000KHz->672000KHz ok!

 

 

Currently it is:

 

 

... and 1 minute later:

 

 

 

The same happened with my Orange Pi+ 2. Orange Pi PC was always running without fan and always is on 624 MHz level:

 

 

 

[tkaiser]

Thank you for the report. Fixed (strangely it was only wrong on those two boards you own).

 

Explanation: The DRAM driver in legacy kernel implements some throttling code that is invoked as soon as cooling state exceeds 0 (in your case that happened when SoC temperature reached 75"C). So the board starts with DRAM clockspeed set by u-boot (624 MHz) but as soon as throttling occurs starts to use the (wrong) value in fex file. Is fixed now and will be rolled out with next major update.

 

BTW: In case H3 doesn't wear a heatsink I would recommend installing one on each board.

 

 

[piknew]

Thanks and thank for the fix . I am using cheap heatsink from aliexpress (do you think I shall put different heatsink or revert to using fan?) - which under heavy load gives (cpuburn-a7 was started at "13:09:02", finished at "13:10:39"):

admin@PKSERVER:~$ sudo armbianmonitor -m
Stop monitoring using [ctrl]-[c]
Time        CPU    load %cpu %sys %usr %nice %io %irq   CPU
13:08:31: 1248MHz  1.04  25%   2%  22%   0%   0%   0%   59°C
13:08:36: 1008MHz  1.04  25%   2%  22%   0%   0%   0%   60°C
13:08:42: 1008MHz  1.04  29%   2%  23%   0%   1%   1%   60°C
13:08:47: 1296MHz  1.03  28%   3%  22%   0%   1%   0%   59°C
13:08:52: 1296MHz  1.03  28%   3%  22%   0%   1%   0%   60°C
13:08:57: 1200MHz  1.03  26%   2%  23%   0%   0%   0%   62°C
13:09:02: 1200MHz  1.43  26%   2%  23%   0%   0%   0%   76°C
13:09:07: 1200MHz  1.71  26%   2%  23%   0%   0%   0%   78°C
13:09:12: 1200MHz  1.98  66%   2%  62%   0%   1%   0%   78°C
13:09:17: 1200MHz  2.22  99%   2%  96%   0%   0%   0%   80°C
13:09:22: 1200MHz  2.44  99%   2%  95%   0%   0%   0%   81°C
13:09:27: 1200MHz  2.65  99%   2%  95%   0%   0%   0%   81°C
13:09:33: 1200MHz  2.83  98%   3%  94%   0%   0%   0%   79°C
13:09:38: 1200MHz  3.01  98%   3%  94%   0%   0%   0%   82°C
Time        CPU    load %cpu %sys %usr %nice %io %irq   CPU
13:09:43: 1008MHz  3.17  98%   3%  94%   0%   0%   0%   83°C
13:09:48: 1008MHz  3.31  99%   2%  96%   0%   0%   0%   83°C
13:09:53: 1008MHz  3.45  99%   3%  94%   0%   0%   0%   77°C
13:09:58: 1200MHz  3.57  99%   3%  94%   0%   0%   0%   79°C
13:10:03: 1200MHz  3.77  99%   3%  94%   0%   0%   0%   83°C
13:10:08: 1200MHz  3.87  99%   3%  94%   0%   0%   0%   79°C
13:10:14: 1008MHz  4.04  99%   3%  94%   0%   0%   0%   80°C
13:10:19: 1008MHz  4.11  99%   2%  95%   0%   0%   0%   80°C
13:10:24: 1008MHz  4.35  98%   4%  93%   0%   0%   0%   83°C
13:10:29: 1008MHz  4.40  98%   4%  93%   0%   0%   0%   80°C
13:10:34: 1008MHz  4.53  98%   3%  93%   0%   0%   1%   81°C
13:10:39: 1296MHz  4.56  98%   3%  93%   0%   0%   1%   69°C
13:10:44: 1008MHz  4.28  98%   3%  93%   0%   0%   1%   68°C
13:10:49: 1008MHz  4.02  55%   2%  51%   0%   0%   0%   67°C
13:10:55: 1008MHz  3.77  28%   3%  21%   0%   1%   0%   67°C
Time        CPU    load %cpu %sys %usr %nice %io %irq   CPU
13:11:00: 1056MHz  3.55  28%   3%  21%   0%   1%   0%   65°C
13:11:05: 1008MHz  3.43  29%   3%  23%   0%   1%   0%   64°C
13:11:10: 1008MHz  3.23  29%   3%  23%   0%   1%   0%   63°C
13:11:15: 1008MHz  3.05  29%   3%  23%   0%   1%   0%   63°C
13:11:20: 1008MHz  2.97  26%   3%  21%   0%   1%   0%   62°C
13:11:25: 1296MHz  2.89  25%   3%  18%   0%   2%   0%   63°C
13:11:30: 1200MHz  2.74  25%   3%  18%   0%   2%   0%   66°C
13:11:35: 1296MHz  2.60  27%   2%  23%   0%   0%   0%   66°C
13:11:40: 1296MHz  2.47  27%   2%  23%   0%   0%   0%   67°C^Cadmin@PKSERVER:~$

 

 

Following confirms as you have mentioned:

 

1. Overall temperature graph by time (this one peak above 60 degrees is rather coincidence, maybe my play with fan connectors ) - in general the highest temperature does not exceed 55-57 degrees).

 

 

2. This is before removing fan (the highest temperature is ~57):

 

 

3. This is after removing fan (the highest temperature is above 70 degrees - however, the logic of ddrfreq seems to be set with lower threshold: around 65 degrees?):

 

 

[tkaiser]

8 minutes ago, piknew said:

however, the logic of ddrfreq seems to be set with lower threshold: around 65 degrees?

 

The throttling code uses the first thermal trip point defined and that's 75°C in default fex file. When you click around in RPi-Monitor graphs using your setting (14 days) will show averaged numbers so for any such 'in depth' looks switching to default graph display (24 hours) is required.

[piknew]

Yes. You are right - this points (624->672) are really on ~75+ level of SoC Temperature.

 

Meanwhile I have changed fex file for both of boards (OPI+2e and OPI+2) as stated in the fix. Rebooted - I will let know if there are any issues (should not be as fix is quite clear and obvious).

[piknew]

Everything is OK.