2 2
Letterus

/proc/cpuinfo show only .LITLLE cores on Odroid-HC1 - hints appreciated

Recommended Posts

Hello,

 

I set up an Odroid-HC1 running Armbian which works fine so far. But today I recognized htop only shows 8 CPUs of type .LITTLE running at a max of 1300Mhz while 4 of them were shown as .BIG CPUs with higher frequency before.

 

The output of cat /proc/cpuinfo is:

 

processor	: 0
model name	: ARMv7 Processor rev 3 (v7l)
BogoMIPS	: 36.00
Features	: half thumb fastmult vfp edsp neon vfpv3 tls vfpv4 idiva idivt vfpd32 lpae 
CPU implementer	: 0x41
CPU architecture: 7
CPU variant	: 0x0
CPU part	: 0xc07
CPU revision	: 3

processor	: 1
model name	: ARMv7 Processor rev 3 (v7l)
BogoMIPS	: 36.00
Features	: half thumb fastmult vfp edsp neon vfpv3 tls vfpv4 idiva idivt vfpd32 lpae 
CPU implementer	: 0x41
CPU architecture: 7
CPU variant	: 0x0
CPU part	: 0xc07
CPU revision	: 3

processor	: 2
model name	: ARMv7 Processor rev 3 (v7l)
BogoMIPS	: 36.00
Features	: half thumb fastmult vfp edsp neon vfpv3 tls vfpv4 idiva idivt vfpd32 lpae 
CPU implementer	: 0x41
CPU architecture: 7
CPU variant	: 0x0
CPU part	: 0xc07
CPU revision	: 3

processor	: 3
model name	: ARMv7 Processor rev 3 (v7l)
BogoMIPS	: 36.00
Features	: half thumb fastmult vfp edsp neon vfpv3 tls vfpv4 idiva idivt vfpd32 lpae 
CPU implementer	: 0x41
CPU architecture: 7
CPU variant	: 0x0
CPU part	: 0xc07
CPU revision	: 3

processor	: 4
model name	: ARMv7 Processor rev 3 (v7l)
BogoMIPS	: 36.00
Features	: half thumb fastmult vfp edsp neon vfpv3 tls vfpv4 idiva idivt vfpd32 lpae 
CPU implementer	: 0x41
CPU architecture: 7
CPU variant	: 0x2
CPU part	: 0xc0f
CPU revision	: 3

processor	: 5
model name	: ARMv7 Processor rev 3 (v7l)
BogoMIPS	: 36.00
Features	: half thumb fastmult vfp edsp neon vfpv3 tls vfpv4 idiva idivt vfpd32 lpae 
CPU implementer	: 0x41
CPU architecture: 7
CPU variant	: 0x2
CPU part	: 0xc0f
CPU revision	: 3

processor	: 6
model name	: ARMv7 Processor rev 3 (v7l)
BogoMIPS	: 36.00
Features	: half thumb fastmult vfp edsp neon vfpv3 tls vfpv4 idiva idivt vfpd32 lpae 
CPU implementer	: 0x41
CPU architecture: 7
CPU variant	: 0x2
CPU part	: 0xc0f
CPU revision	: 3

processor	: 7
model name	: ARMv7 Processor rev 3 (v7l)
BogoMIPS	: 36.00
Features	: half thumb fastmult vfp edsp neon vfpv3 tls vfpv4 idiva idivt vfpd32 lpae 
CPU implementer	: 0x41
CPU architecture: 7
CPU variant	: 0x2
CPU part	: 0xc0f
CPU revision	: 3

Hardware	: SAMSUNG EXYNOS (Flattened Device Tree)
Revision	: 0000
Serial		: 0000000000000000

Any idea on how to make the system detect the CPU cores correctly?

 

Update: Kernel version is 4.19.5-odroidxu4 #5.67 SMP PREEMPT Fri Nov 30 03:15:28 UTC 2018 armv7l GNU/Linux

 

Update: armbianmonitor shows the correct CPU information:

 

### Current system health:

Time       big.LITTLE   load %cpu %sys %usr %nice %io %irq   CPU  C.St.
00:13:18: 1800/ 600MHz  0.06   1%   0%   0%   0%   0%   0% 39.0°C  0/11
00:13:19: 1800/ 600MHz  0.06   4%   3%   0%   0%   0%   0% 38.0°C  0/11
00:13:19: 1800/ 600MHz  0.06   3%   2%   0%   0%   0%   0% 37.0°C  0/11
00:13:20:  600/ 600MHz  0.06   4%   4%   0%   0%   0%   0% 36.0°C  0/11
00:13:20:  600/ 600MHz  0.06   5%   4%   0%   0%   0%   0% 36.0°C  0/11
00:13:24: 1800/1300MHz  0.13   1%   0%   0%   0%   0%   0% 49.0°C  0/11
00:13:24: 1800/1300MHz  0.13   3%   2%   0%   0%   0%   0% 44.0°C  0/11
00:13:24: 1800/1300MHz  0.13   3%   2%   0%   0%   0%   0% 42.0°C  0/11
00:13:25: 1800/1300MHz  0.13   2%   2%   0%   0%   0%   0% 41.0°C  0/11
00:13:25: 1800/1300MHz  0.13   3%   2%   0%   0%   0%   0% 41.0°C  0/11

 

 

Thanks in advance,

Letterus

Share this post


Link to post
Share on other sites
On 1/11/2019 at 1:05 AM, Letterus said:

Any idea on how to make the system detect the CPU cores correctly?

Try using 4.14.69.

Personally, I have not noticed any problems with HC1 and 4.14.69 lately. However, on 4.19.14 I have problems with clocking ... I am not sure how the situation is up to 4.19.5. But according to Igor 4.19.y is not polished for XU4 (HC1 / 2).
And Igor recommends returning to 4.14.y.

Personally, I am currently using 4.14.69 for HC1 without problems, so maybe you can try and let us know how the situation looks like.

Exynos5422 should be clocked up to 2000Mhz for big and 1500Mhz for LITTLE. 1800/1300 is a reduced bar on 4.19.y

Share this post


Link to post
Share on other sites

Thanks for your replies! I tried using 4.14, but it crashed my system. My root dir is on SSD and I'm using a installation of NextcloudPi, maybe this setup is somehow dependend on 4.19.

 

It's htop 2.1.1 and there seems to be a bug in htop indeed. This commands shows speeds up to 1,8Ghz for 4 cores:

watch -n 1  cat /sys/devices/system/cpu/cpu*/cpufreq/cpuinfo_cur_freq

 

And this seems to fit as well:

cat /sys/devices/system/cpu/cpu*/cpufreq/scaling_max_freq 
1300000
1300000
1300000
1300000
1800000
1800000
1800000
1800000

I found this in the output of dmesg, maybe there's a useful hint on what misleads htop?

[    0.272533] CPU: WARNING: CPU(s) started in wrong/inconsistent modes (primary CPU mode 0x1a)
[    0.272538] CPU: This may indicate a broken bootloader or firmware.

Update: cpufreq-info shows correct information as well.

Share this post


Link to post
Share on other sites
18 hours ago, Letterus said:

Thanks for your replies! I tried using 4.14, but it crashed my system. My root dir is on SSD and I'm using a installation of NextcloudPi, maybe this setup is somehow dependend on 4.19.

 

It's htop 2.1.1 and there seems to be a bug in htop indeed. This commands shows speeds up to 1,8Ghz for 4 cores:


watch -n 1  cat /sys/devices/system/cpu/cpu*/cpufreq/cpuinfo_cur_freq

 

And this seems to fit as well:


cat /sys/devices/system/cpu/cpu*/cpufreq/scaling_max_freq 
1300000
1300000
1300000
1300000
1800000
1800000
1800000
1800000

I found this in the output of dmesg, maybe there's a useful hint on what misleads htop?


[    0.272533] CPU: WARNING: CPU(s) started in wrong/inconsistent modes (primary CPU mode 0x1a)
[    0.272538] CPU: This may indicate a broken bootloader or firmware.

Update: cpufreq-info shows correct information as well.

I'm not sure where the problem is.

Your Exynos5422 is detected almost as it should. Only 1800/1300 is bad but it's the fault of 4.19.
Are you sure that htop simply does not give you the same cores twice?
In my case, htop 2.1.1 on both 4.14 and 4.19 gave the correct reading(2000/1500 - 1800/1300). Make sure that you have the cores properly set in htop. And if you stay on 4.19 remember that you lose 200Mhz on the core.

 

 

hc1.jpg

Share this post


Link to post
Share on other sites

jshc1: I've no problem with loosing 200Mhz (the cooling plate still gets quite hot sometimes), but I would have if the system uses the wrong cores for the hard work.

 

Thanks to your hint I detected htop was set up wrong. I don't know how this happened, but hinstead of showing CpuFreq1 to 8 it was showing the variable CpuFreq (without any number) 8 times. CpuFreq only showed the "little" CPU freq.

 

So, thanks to this hint, I consider this issue solved. :-)

Share this post


Link to post
Share on other sites
1 hour ago, Letterus said:

jshc1: I've no problem with loosing 200Mhz (the cooling plate still gets quite hot sometimes), but I would have if the system uses the wrong cores for the hard work.

 

Thanks to your hint I detected htop was set up wrong. I don't know how this happened, but hinstead of showing CpuFreq1 to 8 it was showing the variable CpuFreq (without any number) 8 times. CpuFreq only showed the "little" CPU freq.

 

So, thanks to this hint, I consider this issue solved. :-)

 

It's good that the problem is solved. :)



If you have any stubborn program, you can always let it run on specific cores. I'm doing it with Kodi 17.1. This mofo is so greedy on the cores that eats all 8 in 90% regardless of whatever. It just so happens that I use kodi as a poor audio player and I only need to give it three litlle cores to perform its task. it does not occupy big cores, which at the same time allows them to sit on 600Mhz and in theory less heat and electricity.

Share this post


Link to post
Share on other sites
2 2