S802 - Truth, Lies, and Video Tape - AML needs to answer to consumers

[Finless]

It's time to let you know something I found recently.

NOW.... no one has told me this. This is a observation I have made having S802 boxes from the beginning and a brand new box from Minix.

Being a "hacker dev" I use dmesg and logcat to see what's going on and do my work to fix things in ROMs when I can. I pull apart things and work on .smali code in the ROM.
Right now, kernel patching is impossible on S802 due to the fact it is CRC signed. I have not figured how to crack that CRC yet for S802 but for a long time we have been doing exactly that on RockChip kernels well before we had sources leaked.
So patching a kernel binary is a no go... But it's something I like to do rather than working from SCRATCH on sources. In my opinion, that's the long road and hard way. BUT it's just my personal preference because I have been doing it that way for so long.... You know, dog and new tricks

I am not a work from sources like wasser and others. Not that I can't, I just prefer to reverse engineer and see what's going on.




So I have spotted this recently!

It's a DMESG log that shows something has changed.

AML has changed their processors! Not saying it's better and not saying it's bad! Just saying there is a DIFFERENCE! This has to do with thermal throttling of the CPU.

If I pull a dmesg log from one of my early boxes it says this:

amlogic_thermal_probe, this chip is not trimmed, use virtual thermal
aml_virtaul_thermal_probe, virtual thermal enabled

Of note here is the "virtual" thermal!

Now on a new box I just got, this is what dmesg says:

amlogic_thermal_probe, this chip is trimmed, use thermal
amlogic-thermal aml_thermal: amlogic thermal probe start

Notice the difference?

So AML has changed something! Not sure what it is but my new box I just got runs cooler with far less CPU throttling.

I am telling you this because it's the right thing to do. AML can bite me if they ask me to pull this post! It's a FACT JACK! Something has changed and I think it is for the better that people know about this!
At least RockChip identified the 3188T visually rather than not do anything and "sneak it in"...

NOW... DO NOT GO beating on Geekbuying, Beelink, Minix, or anyone else for a refund! They probably did NOT KNOW about this issue from the start! It is unfair to them to ask for a refund! It's not their fault! They just bought the processor, put it on a board, and sold it!!! You cannot blame them!

AML is responsible for explaining to customers what this is all about! If your upset or concerned, contact AML. In my opinion, maybe the processor was released to early and now some are going to be stuck in the past! New processors have a difference (good or bad I don't really know) and that's the truth.

It's only fair for me to tell everyone what I am seeing and show the information as such.
FreakTab is about truth, honesty, and good ROMs

Bob

[pigbait]

Thanks Bob for all your work.

[gizmomelb]

nice observation Bob, let's see what response Amlogic put forward.

[TMx]

Minix was right about the cpu stuff but on the other hand We as a Square M8 owners never experienced such overheating issues at all...

[Finless]

Minix was right about the cpu stuff but on the other hand We as a Square M8 users never experienced the overheating issues at all...

My ENY M8 and both my S89's have this. But they are not over heating!
They do what I ask them to do and run what's is needed! How the heck have I been making ROMs and using the boxes? So not sure this is a real issue other than a change!

Again, I am not sure all this means ANYTHING in the end. Only that there is a difference with the CPU's now.

I know that some think this is an issue. I saw it early on... but the boxes run and do everything I test and do! Not sure what else to say other than early CPUs are now obsolete in some way and new proc's will come with this new configuration.

I expect this to happen again BTW... It did with many other SOC's over time.

Bob

[rlaursen]

My ENY M8 and both my S89's have this. But they are not over heating!
They do what I ask them to do and run what's is needed! How the heck have I been making ROMs and using the boxes? So not sure this is a real issue other than a change!

Again, I am not sure all this means ANYTHING in the end. Only that there is a difference with the CPU's now.

I know that some think this is an issue. I saw it early on... but the boxes run and do everything I test and do! Not sure what else to say other than early CPUs are now obsolete in some way and new proc's will come with this new configuration.

I expect this to happen again BTW... It did with many other SOC's over time.

Bob

I've noticed when importing games to rom browser in xbmc.. after about 15 or 20 minutes the box gets real hot and completely freezes up with the EM8 but thats maxing out the cpu for a long time doing something you dont do often. These boxes WILL overheat if you max them out for a period of time and require a reboot. Makes me wonder if that was why they initially clocked them down to 1.6ghz

[pigbait]

BOB any way of knowing what cpu's are what yet?

[Finless]

BOB any way of knowing what cpu's are what yet?

In the first post I TOLD YOU what to check.

I know some people may not be technical enough to do that, but sorry, that is as far as I know how to check.

Done....

Bob

[pigbait]

Thanks....

[Batfink32]

"Trimming" im sure has something to do with temperatures. The trimmed chips run cooler than the non trimmed chips. This is what i think is happening on the x8 boxes also,there seems to be temperature differences of around 20degc. I spoke to another poster yesterday who was getting temps of 40degC idle on his X8. He looked at his DMESG and it says his chip was trimmed. My X8 runs idle at 55degC and i cant see any "Trimmed" in my DMESG. (note these are the "real" temps from the beta 2 rom.)

[no_spam_for_me]

trimmed vs. not trimmed OR 'use thermal' vs. 'use virtual thermal'

.
trimmed means, that the SoC is able to determined real and truths temperatures and because of that, the kernel can use this value to "manage" the SoC (throttling the max freq and using cores).

not trimmed is the opposite of it, so the kernel must find a way to "manage" the SoC... AND at one of my devices (s89 with not trimmed SoC) it means that it seams so, that they use the SoC usage and/or freq to calculate thermal value (virtual)...
I find it out at my stress test at e.g. 102k4 and 104k4 and see the effect, that the thermal value jumps at minimum 5°C steps and if I use e.g. "816/1992 interactive" I got at idle > 60°C and at stock 40/45°C (and that's why I call it 'fake')...

[no_spam_for_me]

... who was getting temps of 40degC idle on his X8. He looked at his DMESG and it says his chip was trimmed. My X8 runs idle at 55degC and i cant see any "Trimmed" in my DMESG. (note these are the "real" temps from the beta 2 rom.)

I get a little bit higher values, but I see, that it is trimmed:

Idle at stock 20140529 with starting at 42/43°C and after 5 minutes at 47/48°C.
Idle at self pached interactive 816/1.992 stock 20140529 also the same.

adb shell dmesg > dmesg.txt

searching at dmesg.txt for 'trimmed' =>

<4>[    5.450041@3] efuse_flag=a
<4>[    5.452729@3] amlogic_thermal_probe, this chip is trimmed, use thermal
<6>[    5.459239@3] amlogic-thermal aml_thermal: amlogic thermal probe start
<4>[    5.465856@3] #thermal-cells=7

BTW: If you search case sensitive you aren't able to find it with 'Trimmed' ...

[Jimandroidpc]

Laymans term: is it actually a new chip?

[Batfink32]

Re: S802 - Truth, Lies, and Video Tape - AML needs to answer to consumers

I get a little bit higher values, but I see, that it is trimmed:

Idle at stock 20140529 with starting at 42/43°C and after 5 minutes at 47/48°C.
Idle at self pached interactive 816/1.992 stock 20140529 also the same.

[ATTACH=CONFIG]6898[/ATTACH]

adb shell dmesg > dmesg.txt

searching at dmesg.txt for 'trimmed' =>

<4>[    5.450041@3] efuse_flag=a
<4>[    5.452729@3] amlogic_thermal_probe, this chip is trimmed, use thermal
<6>[    5.459239@3] amlogic-thermal aml_thermal: amlogic thermal probe start
<4>[    5.465856@3] #thermal-cells=7

BTW: If you search case sensitive you aren't able to find it with 'Trimmed' ...

I've ran about 6 dmesgs now and never found "Trimmed" . Idle with the beta 2 I'm up at 55. It must be the difference between trimmed and not trimmed.

Sent From My LG G3 Using Tapatalk

[no_spam_for_me]

Maybe try to use stock 05/29 and have a look at dmesg...
(maybe they don't give a kernel message if not trimmed or no message at beta???)

I do really quick after booting a dmesg (150KByte) and also later (310KByte)
and at both I find it (but only with 'trimmed' and not 'Trimmed' )

dmesg output begin with
<6>[ 0.000000@0] Booting Linux on physical CPU 0x200

(at the 150KByte nearly at the 1/2 I find this information...)