Armbian for Amlogic S912
25 25

524 posts in this topic

Recommended Posts

19 minutes ago, jeanrhum said:

I haven't opened my gt1 ultimate.

 

Thank you for testing! I would keep the box closed since it shows excellent heat dissipation (25"C less than VIM2 with same load -- that's really impressive). Out of curiousity: what material is the enclosure made of and does it feel warm(er)?

 

@balbes150: Do you have a pillow at home to throw on the VIM2? ;) I can't understand that no one is able to get the S912 throttling! :) 

Share this post


Link to post
Share on other sites
what material is the enclosure made of and does it feel warm(er)? 

The enclosure is in platic. I will look closer to the enclosure and its temperature next time. Initial temperatures were around 36°C.

I will also try to take more time to run this test, because I think the temperature could have continue to slowly increase but it was too late and I didn't want to let it run like that during night. I also have a H96 pro+ and the gt1 ultimate is much more heavy. I suppose that there is a big heatsink over the soc whereas h96pro+ has nothing (but I didn't succeed in running armbian on it, I suppose due to reboot issues even if I manage to install latest superceleron's rom from freaktab).

Share this post


Link to post
Share on other sites
18 minutes ago, jeanrhum said:

I will also try to take more time to run this test

 

Usually 15 minutes should be sufficient (at least that's my experience). Wrt Beelink B1 heat dissipation forget my question please. Jean-Luc already provided insights over a year ago: https://www.cnx-software.com/2016/09/09/beelink-gt1-4k-tv-box-review-part-1-unboxing-and-teardown/

 

So the heat gets spread to this large metal plate at the enclosure bottom which seems to be rather efficient. :) 

Share this post


Link to post
Share on other sites

Wrapped in a warm hat is one of my TV boxes (Tronsmart Vega S96) :) .

Run test for a couple of hours. Here is the result at the end of the second hour. Such a conclusion in the console, no change for end hours, so stopped the test.

 

Spoiler

1512 @ 79:     execution time (avg/stddev):   4.1436/0.01
1200 @ 79:     execution time (avg/stddev):   4.0925/0.01
1512 @ 79:     execution time (avg/stddev):   4.1210/0.00
1200 @ 79:     execution time (avg/stddev):   4.1082/0.01
1512 @ 79:     execution time (avg/stddev):   4.0899/0.01
1200 @ 79:     execution time (avg/stddev):   4.1200/0.01
1200 @ 79:     execution time (avg/stddev):   4.0957/0.01
1200 @ 79:     execution time (avg/stddev):   4.1305/0.01
1512 @ 79:     execution time (avg/stddev):   4.1274/0.01
1200 @ 79:     execution time (avg/stddev):   4.0736/0.01
1200 @ 79:     execution time (avg/stddev):   4.1054/0.01
1200 @ 79:     execution time (avg/stddev):   4.1650/0.01
1512 @ 79:     execution time (avg/stddev):   4.0857/0.01
1512 @ 79:     execution time (avg/stddev):   4.0908/0.01
1512 @ 79:     execution time (avg/stddev):   4.1230/0.01
1512 @ 79:     execution time (avg/stddev):   4.1660/0.00
1200 @ 79:     execution time (avg/stddev):   4.0707/0.00
1200 @ 79:     execution time (avg/stddev):   4.1134/0.01
1512 @ 79:     execution time (avg/stddev):   4.1198/0.01
1200 @ 79:     execution time (avg/stddev):   4.0714/0.01
1200 @ 79:     execution time (avg/stddev):   4.1465/0.01
1512 @ 80:     execution time (avg/stddev):   4.0923/0.01
1200 @ 79:     execution time (avg/stddev):   4.0756/0.01
1200 @ 79:     execution time (avg/stddev):   4.2131/0.01
1200 @ 79:     execution time (avg/stddev):   4.1439/0.00
1512 @ 79:     execution time (avg/stddev):   4.0813/0.01
1512 @ 79:     execution time (avg/stddev):   4.1081/0.01
1512 @ 79:     execution time (avg/stddev):   4.1305/0.01
1512 @ 79:     execution time (avg/stddev):   4.0961/0.00
1200 @ 79:     execution time (avg/stddev):   4.1359/0.01
1512 @ 79:     execution time (avg/stddev):   4.1054/0.01
1512 @ 79:     execution time (avg/stddev):   4.0900/0.01
1512 @ 79:     execution time (avg/stddev):   4.0790/0.01
1512 @ 79:     execution time (avg/stddev):   4.0888/0.01
1200 @ 79:     execution time (avg/stddev):   4.0865/0.01
1200 @ 79:     execution time (avg/stddev):   4.1416/0.01
1512 @ 79:     execution time (avg/stddev):   4.1541/0.01
1512 @ 79:     execution time (avg/stddev):   4.0418/0.01
1512 @ 80:     execution time (avg/stddev):   4.1571/0.01
1512 @ 79:     execution time (avg/stddev):   4.1074/0.01
1512 @ 80:     execution time (avg/stddev):   4.1449/0.01
1200 @ 79:     execution time (avg/stddev):   4.1127/0.01
1200 @ 79:     execution time (avg/stddev):   4.1390/0.01
1200 @ 79:     execution time (avg/stddev):   4.1569/0.01
1512 @ 79:     execution time (avg/stddev):   4.1365/0.01
1200 @ 79:     execution time (avg/stddev):   4.0810/0.01
1512 @ 79:     execution time (avg/stddev):   4.1297/0.01
1200 @ 79:     execution time (avg/stddev):   4.0858/0.01
1512 @ 79:     execution time (avg/stddev):   4.0995/0.01
1200 @ 79:     execution time (avg/stddev):   4.1034/0.01
1512 @ 79:     execution time (avg/stddev):   4.0843/0.01
1512 @ 79:     execution time (avg/stddev):   4.1451/0.01
1200 @ 79:     execution time (avg/stddev):   4.1349/0.01
1512 @ 79:     execution time (avg/stddev):   4.1799/0.01
1512 @ 79:     execution time (avg/stddev):   4.1290/0.01
1200 @ 79:     execution time (avg/stddev):   4.2215/0.01
1200 @ 79:     execution time (avg/stddev):   4.1333/0.01
1512 @ 79:     execution time (avg/stddev):   4.1494/0.01
1200 @ 79:     execution time (avg/stddev):   4.1901/0.01
1512 @ 79:     execution time (avg/stddev):   4.1292/0.01
1512 @ 79:     execution time (avg/stddev):   4.1551/0.01
1200 @ 79:     execution time (avg/stddev):   4.2086/0.01
1512 @ 79:     execution time (avg/stddev):   4.1498/0.01

 

willmore and tkaiser like this

Share this post


Link to post
Share on other sites
14 hours ago, balbes150 said:

1200 @ 79:     execution time (avg/stddev):   4.2215/0.01

 

Thank you for the test. So this box starts to throttle around 80°C and reported cpufreq also decreases (when 1512 is shown in reality it's 1200 MHz, the above execution duration indicates ~1080MHz in reality). So numbers still bogus as expected but at least there is an indication that throttling happens through sysfs (Raspberry Pi for example fakes this also: when throttling happens there and the real cpufreq goes below 1200 MHz -- eg. 800 MHz -- sysfs still reports 1200 MHz until 600.1 MHz. Only if throttled down to 600.0 MHz sysfs shows correct values again).

Share this post


Link to post
Share on other sites

Quoting a certain popular S912 ROM's Changelog:

 

Quote

Added True 8 core ( All cores now work at 1.5ghz )

Does anyone know if this feature could be ported to the Armbian kernel? I'm sure it would solve those weird ramspeed results reported earlier.

Share this post


Link to post
Share on other sites
8 minutes ago, Shimon said:

Does anyone know if this feature could be ported to the Armbian kernel? I'm sure it would solve those weird ramspeed results reported earlier.

 

Ramspeed? Kernel?

 

The issue is that the firmware blob is lying to the kernel wrt cpufreq (not RAM -- testing with sysbench is excluding RAM 100%). How should something in the kernel fix firmware behaviour? What's the proof of 'All cores now work at 1.5ghz'? The fantasy values returned by /sys/devices/system/cpu/cpu0/cpufreq/scaling_cur_freq?

Share this post


Link to post
Share on other sites
27 minutes ago, tkaiser said:

 

Ramspeed? Kernel?

Please, read the thread before spouting question marks. I know what the issue is, I originally benchmarked it. The kernel is almost certainly involved in that observed performance drop due to big.little scheduling. 

 

Superceleron's mod is definitely worth investigating.

Share this post


Link to post
Share on other sites
24 minutes ago, Shimon said:

read the thread

 

Of course not. If you can post links to interesting posts somewhere in between this 172 pages (!) thread or a github repo with the changes I'll take a look ASAP. Thanks.

Share this post


Link to post
Share on other sites

How do i get my H96Pro+ to work with a DVI monitor (it works fine with our Samsung TV, HDMI input)? I've tried to with HDMI-to-DVI adapter (this could be opposite adapter not sure), no signals on the monitor. 

Do i need active HDMI-to-DVI adapter?

 

Will any Orange Pi H3 or H5 boards have the same issues too? I'm planning to buy an Orange Pi board that has accelerated X11/GUI so that i can use it as a simple desktop, mainly to access my KVM VM over SPICE protocol.

 

Share this post


Link to post
Share on other sites

Maybe it is only for Allwinner SoCs, but you can try (Pine64 doc):

If you use a DVI display don’t forget to define disp_dvi_compat=1 in /boot/armbianEnv.txt (supported starting with 5.21).

 

Share this post


Link to post
Share on other sites

Data for the Odroid-C2 for 2 threads to fill out the table @tkaiser.

100:     execution time (avg/stddev):   196.7230/0.01
250:     execution time (avg/stddev):   76.9758/0.01
500:     execution time (avg/stddev):   37.5997/0.00
1000:     execution time (avg/stddev):   18.3120/0.00
1296:     execution time (avg/stddev):   14.3056/0.00
1536:     execution time (avg/stddev):   12.0588/0.00
1656:     execution time (avg/stddev):   11.1768/0.00
1680:     execution time (avg/stddev):   11.0148/0.00
1752:     execution time (avg/stddev):   10.5501/0.00

 

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

25 25

  • Support the project

    We need your help to stay focused on the project.

    Choose the amount and currency you would like to donate in below.