Stress Test Scenario

[no_spam_for_me]

Beforehand
every tool I like to monitor the system seems to be able to work correctly at this system (I know there exists also newer versions )
'Cpu Temperature' v1.5

https://play.google.com/store/apps/details?id=com.yanyan.cputemp

'CPU Monitor' v5.8

https://play.google.com/store/apps/details?id=com.bigbro.ProcessProfiler

'System Tuner Pro' v3.0.9

https://play.google.com/store/apps/details?id=ccc71.pmw.pro

But I'm not sure about this tool because I can't get 100% CPU utilisation, maybe because of big.LITTLE handling (by the Android-system)?
'StabilityTest (ROOT optional)' v2.7 => classic test

https://play.google.com/store/apps/details?id=com.into.stability

First shot
This SoC is also temperature sensitive like the amlogic S802 http://www.freaktab.com/showthread.p...-Test-Scenario (instead of the RK3288 http://www.freaktab.com/showthread.p...-Test-Scenario ) and early begins to throttle ...

http://oi59.tinypic.com/mr8rjr.jpg

---

At CPU utilisation/usage all the tools seems to have problems to calculate the right values at the moment.
E.g. 'Cpu Temperature' v2.0 shows at peak 285%

http://oi58.tinypic.com/2mr9ts7.jpg

But at the main part it is useful because I think the temperature value is correct (btw: the curve is also running/monitoring/determining if the app is running at background)

---

Interactive

At this governor 1 or more LITTLE cores (A7) a often offline (I'm using 'CPU Monitor' v5.92 to see it at a curve):

http://oi62.tinypic.com/333jwa1.jpg

Here is the monitoring

Throttling
> 60°C -> 1.6 GHz at big (reached really fast / nearly directly)
> 75°C -> 1.41 GHz at big (maybe after ~7 minutes)
> 85°C -> 1.2 GHz at big (maybe after ~20 minutes (BUT not really sure), but at the beginning most at 1.41 and at the end of the monitoring (after 60 minutes) 50:50 jumping between 12 / 1.41 GHz)

What I don't understand is the much different CPU Load (usage/utilisation) but maybe all these tools at the moment not 100% optimized to this SoC???

Also I don't know what I have to think about the way how they/it determine the temperature value.
If you have a look at the 5 min cool-down you can see it decreases in order to ~13 °C nearly at 1-2 sec and after that maybe ~10 °C during the next ~10 sec, so I can't belief that they work with hw-sensors but they calculate it ... BUT at the other hand the big heatsink and the alu-case... (also if I have a look at the first 5 min stress-test)???


3 min idle

http://oi60.tinypic.com/r0zih4.jpg

5 min stress-test

http://oi59.tinypic.com/2hx59as.jpg

10 min stress-test

http://oi57.tinypic.com/dyorc4.jpg

~30 min stress-test (not easy to catch the right moment to get 1.2 GHz at big )

http://oi60.tinypic.com/9h4zly.jpg

60 min stress-test

http://oi59.tinypic.com/29qdxg5.jpg

5 min cool down

http://oi59.tinypic.com/2nqutkx.jpg

[no_spam_for_me]

UPDATE post #1

[no_spam_for_me]

Governor

Available governor
- interactive (default)
- ondemand
- userspace
- performance

Here I find a nice explanation of the different governor types:

http://androidforums.com/threads/android-cpu-governors-explained.513426/

(I delete the parts which belongs to the battery life )

OnDemand
This governor has a hair trigger for boosting clockspeed to the maximum speed set by the user. If the CPU load placed by the user abates, the OnDemand governor will slowly step back down through the kernel's frequency steppings until it settles at the lowest possible frequency, or the user executes another task to demand a ramp.

OnDemand has excellent interface fluidity because of its high-frequency bias...

This final fact is important to know before you read about the Interactive governor: OnDemand scales its clockspeed in a work queue context. In other words, once the task that triggered the clockspeed ramp is finished, OnDemand will attempt to move the clockspeed back to minimum. If the user executes another task that triggers OnDemand's ramp, the clockspeed will bounce from minimum to maximum. This can happen especially frequently if the user is multi-tasking.

Interactive
Much like the OnDemand governor, the Interactive governor dynamically scales CPU clockspeed in response to the workload placed on the CPU by the user. This is where the similarities end. Interactive is significantly more responsive than OnDemand, because it's faster at scaling to maximum frequency.

Unlike OnDemand, which you'll recall scales clockspeed in the context of a work queue, Interactive scales the clockspeed over the course of a timer set arbitrarily by the kernel developer. In other words, if an application demands a ramp to maximum clockspeed (by placing 100% load on the CPU), a user can execute another task before the governor starts reducing CPU frequency. This can eliminate the frequency bouncing discussed in the OnDemand section. Because of this timer, Interactive is also better prepared to utilize intermediate clockspeeds that fall between the minimum and maximum CPU frequencies.
:
Interactive also makes the assumption that a user turning the screen on will shortly be followed by the user interacting with some application on their device. Because of this, screen on triggers a ramp to maximum clockspeed, followed by the timer behavior described above.

Userspace
This governor, exceptionally rare for the world of mobile devices, allows any program executed by the user to set the CPU's operating frequency. This governor is more common amongst servers or desktop PCs where an application (like a power profile app) needs privileges to set the CPU clockspeed.

Performance
This locks the phone's CPU at maximum frequency. While this may sound like an ugly idea, there is growing evidence to suggest that running a phone at its maximum frequency at all times will allow a faster race-to-idle. Race-to-idle is the process by which a phone completes a given task, such as syncing email, and returns the CPU to the extremely efficient low-power state. This still requires extensive testing, and a kernel that properly implements a given CPU's C-states (low power states).

Because of this I'm interested in Performance.

BTW: For those how are interested in the 'I/O scheduler' should have a look at http://androidforums.com/threads/cpu...nation.642199/
(Available: noop, deadline and cfq (default) )


AnTuTu CPU Master Free v2.5.3

https://play.google.com/store/apps/details?id=com.antutu.CpuMasterFree

I don't know why min is set to 1.608GHz and the funny thing is, if I can trust 'System Tuner Pro v3.2' (because here I see min set to 1GHz), that this only effect to ONE core...


If I change min to 1.8GHz it's also shown at 'System Tuner Pro v3.2'



SetCPU for Root Users v3.1.2

https://play.google.com/store/apps/details?id=com.mhuang.overclocking

Also only ONE core but the min at 'System Tuner Pro v3.2' is corresponding with the min at 'SetCPU for Root Users v3.1.2' (so I think the issue is at 'AnTuTu CPU Master Free v2.5.3')

No-frills CPU Control v1.28.2

https://play.google.com/store/apps/details?id=it.sineo.android.noFrillsCPU

The same as 'SetCPU for Root Users v3.1.2' BUT [Apply at Boot] doesn't work!


Conclusion
Maybe you can say there exists no 'Multi-core control support' at the kernel.
At idle only the one core (0) is at performance and all other cores are at interactive but also offline, BUT at that moment the kernel needs one of the other cores (e.g. 1,2), it switches form interactive to performance, BUT you can't set each core as performance (and also online) at default...


.

[no_spam_for_me]

Stress test at governor Performance

The temperatures (and of cause the throttling) seems to be the same, BUT, especially during the first 10 minutes, the CPU usage/utilisation is more often at the peak (80-100%) BUT the base value (full-gray at the curve) seems to be higher at interactive after ~ 5 minutes of the stress test (gets continues smaller/lower at performance)...

3 min idle

http://oi57.tinypic.com/dxekq9.jpg

5 min stress-test

http://oi62.tinypic.com/6foa53.jpg

10 min stress-test

http://oi59.tinypic.com/abjt4.jpg

30 min stress-test

http://oi61.tinypic.com/r9o1hu.jpg

60 min stress-test

http://oi58.tinypic.com/5xr815.jpg


5 min cool down
http://oi58.tinypic.com/6givch.jpg

[no_spam_for_me]

UPDATE post #4

[Finless]

Cool stuff!

Bob

[no_spam_for_me]

Maybe governor Performance have also a positive effect at XBMC/Kodi ...
=> http://www.freaktab.com/showthread.p...l=1#post259505

[no_spam_for_me]

A80 vs. RK3288

CPU

I try to find an app, which is able to loop math operations to determine how much the SoC-CPU and it cores are able to calculate during a defined time, but I don't find something like this, but I know the 'StabilityTest - Classic Test' and I hope I interpret the 'successful runs' correct as one cycle ...

http://intostudios.com/stability

	4 minutes	10 minutes
A80	~ 12	~ 29
RK3288	~ 9,6	~ 22,6

RK3288 => UBOX 2/8GB 111k4


4 min
A80

RK3288

(http://oi57.tinypic.com/23koffd.jpg / http://oi59.tinypic.com/2nh3n1t.jpg)


10 min
A80

RK3288

(http://oi57.tinypic.com/sq3447.jpg / http://oi62.tinypic.com/2e5lk5v.jpg)



.

[no_spam_for_me]

Heatsink mod ...

... useful or useless?

I only tried to find out, if some extra cooling will help the system.
I used the pcb without the case and lifted it up at one corner (~1.5 cm) and a 15 cm table fan ~ 15 cm away from the heatsink...

That's the first 5 minutes without these "mods" (the picture from post #4)

http://oi62.tinypic.com/6foa53.jpg



Here we are with the "mods" during the first 5 minutes:
(You can see that I first start the sress test and then the fan )

http://oi60.tinypic.com/5xjrb.jpg


The next 5 minutes I'm using it without the table fan to get higher temperatures so I can have a look at the third 5 minutes again using the table fan:

http://oi57.tinypic.com/9fv5gk.jpg

---

The good news: the mod effects the temperature and also a little bit of the performance (which you can the at the third 5 minutes where the 'CPU Load' gets a little bit higher)...

The 'bad' news: I/you have to ask yourself if/whether this effort is in proportion to the result...

I think a replacement of the thermal pad between the SoC and the heatsink could be an idea, because the standard Chinese thermal pad has worse W/mK (higher is better)... often they are at 1-2 W/mK instead of e.g. >= 5 W/mK... (the original thickness seems to be 2mm, but I think thinner is also possible because of the way the heatsink is mounted, and original length/width is 60mm)...
e.g. http://www.****.de/itm/Phobya-Warmel...item418fe0aeab

Because the underside of the pcb gets also warm, but not really hot, and the space between pcb and the case seems to be less than 5mm (~4.5mm) it could be an idea to use something like this http://www.****.de/itm/EC-360-BLUE-5...item1c3ea70985

Also the heatsink itself isn't the best (original max high is 10mm) but of course better then nothing...
The space above this heatsink to the case is ~ 5-6mm (space between pcb and case is 18mm - heatsink with 2mm thermal pad at pcb is ~ 11.5mm till 12.5mm)...
But because of the design of the heatsink I think it's senseless to use also a 5mm pad at this side...

.

[Boxx]

Lets watercool it!
Just kidding...
Its a pity that it cant stand the 1,8ghz at least. Maybe it would eliminate the micro stutter (more nano stutter now).
Installing a fan is no option for me. I dont want to cut a hole into the housing and i want my box to be quiet.

[no_spam_for_me]

Yes, of course, but because of this

:
edit: ok, now i see why i can´t set it to 1800. cpu is on 59-60 degrees celsius when idle @ 1600. that means, as soon as i switch to 1800, it reaches 60 degrees and then gets throttled back to 1600.
:

you should have a look at the pad and heatsink...
As you could see, at close case I got < 30°C (see e.g. post #4) or do you have 'heavy' processes running at 'idle' or means 'idle' after running processes like e.g. KODI?
EDIT: I use SetCPU...

[Boxx]

Yes, of course, but because of this

you should have a look at the pad and heatsink...
As you could see, at close case I got < 30°C (see e.g. post #4) or do you have 'heavy' processes running at 'idle' or means 'idle' after running processes like e.g. KODI?
EDIT: I use SetCPU...

i got this temp right after booting up, no apps running. the case feels pretty warm too, so i think the values are right. i know its way to hot. i believe it wasn´t that hot with homlet rom, will check that. maybe the tweaks on the new dovless rom are causing the heat...
on the other hand, i have the mk80 with plastic case but i cant believe that this can cause a temp difference of 20-30 degrees. even a low quality thermal pad can´t explain this or do you think different?
i will take a video of whats going on...

[Boxx]

i uninstalled a few apps and the temp went down 3-4 degrees.
now i see the cpu running at 1,8ghz (idle) for the first time since i am on dovless.

i made 2 short videos. one taken with my cell phone, showing cpu manager stats right after booting, no running apps. on the other video, first there is only the screen capture app running. then i run the stress test app.

added: installed homlet to sd card, installed 2 apps, no further changes, no running apps, temp: 54-55 degrees.
makes me believe something is wrong with my device. would be interesting to see some more stats from other rikomagic mk80 boxes.

[no_spam_for_me]

hmmm, I don't know the (new) dovless because I'm at stock rom (http://www.freaktab.com/showthread.p...ock-20141219v1) and the system isn't at 4 + cores at idle (only one core) and also the CPU Usage is much more less (which you can see at my tests)...

[no_spam_for_me]

EDIT: ok, the first video is the capture/stress test video and the scound one is the handy cam video which is private, so can't watch it...