(S922x, S905x3) Odroid N2 vs Khadas VIM 3 vs ...?

  • Mainline-based LE also works on Odroid-N2 and it was ready earlier than for VIM3 simply because the Odroid was ready earlier.

    It is a work in progress but support for both boards should be almost equal in mainline-based builds. I also think that CE support is a matter of creating a device tree for VIM3 and like you said: most of work for S922X was already done for N2 and every S922X can reuse it.

    Is hw decoding working on the N2 with mainline?

    And how about hdmi audio and ethernet?

  • Any reputable company which is prepared to send samples and offer a bit of cooperation will get supported by CE.
    The take home message has to be don't use fly by night no name generic box sellers.

    Shoog

    • Official Post

    I see both HK and Khadas as different shades of the same grey when it comes to software support. HK definitely have a more active user community around their products and some reasonable staff, but you can count the number of upstream kernel/u-boot/etc. commits from those staff on the same lack of fingers as Khadas (and several other SBC vendors). At the end of the day they all have the same business model which relies on seeding free samples among the Linux community in the hope that enthusiasts will do most of the work for them. The *only* board manufacturer who gets my vote for contribution is LibreComputer; who don't code anything themselves, but have been bankrolling Amlogic GX mainline development for some time - which has directly benefitted the entire community in addition to their own products. LibreComputer customers don't need much vendor support because you can just grab the latest upstream source for kernel, u-boot, etc. and everything is already supported.

    VIM3 has a reworked heatsink and fan; there are taller vanes on the heatsink and the fan (which is allegedly silent now) has swapped sides to improve the cooling .. although I wouldn't know as Khadas forgot to include one in the sample kit. I have the board in one of their fancy cases (which I think are a bit crap, but they do the job) and it's warmer than cool (and the N2) but also shows no signs of overheating and it's still nowhere near the "raging inferno" temp that mk1 AppleTV's used to reach. Once G12 devices start running at full speed I might need to check the temps - right now they're on minimum clock at 1.2GHz (and still faster than previous generations) - but then adding the fan is the obvious addition. I wouldn't want to run one for long without the heatsink.

    NB: The initial VIM3 device-tree that I made was literally a 10 mins copy/paste exercise (in a Hotel room .. no special tools needed). Once we have a few more devices under our belt I think G12A/B will be an identikit exercise.

    • Official Post

    Is hw decoding working on the N2 with mainline?

    And how about hdmi audio and ethernet?

    Yes .. with caveats for 4K and HEVC which have dependencies on 10-bit video work (which is now active work-in-progress again). HDMI audio is 2.0 at the moment but we've started looking into what's needed for multi-channel .. although bizarely 7.1 pass-through (AC3/DTS) also appears to be working as long as you set 'optimised' not 'best match' in Kodi. Ethernet is no problem.

  • Yes .. with caveats for 4K and HEVC which have dependencies on 10-bit video work (which is now active work-in-progress again). HDMI audio is 2.0 at the moment but we've started looking into what's needed for multi-channel .. although bizarely 7.1 pass-through (AC3/DTS) also appears to be working as long as you set 'optimised' not 'best match' in Kodi. Ethernet is no problem.

    Great thanks!

  • ... about the poor efficiency of the cooling system on N2 (costs do not meet the requirements of the result), referring to the fact that at that time I had no sample of N2 (by the way, actual tests have confirmed the correctness of my calculations)....

    I must correct you (as others did many times) ... I'm no fanboy but rather manufacturer & SOC agnostic; always looking for the hardware which suits my needs best. So: just an ordinary user and I still have to say that the N2 cooling / temperature curve is superior to anything else I saw ...

    Comparison; both more or less in some sort of idle mode:

    My RockPi4:

    LibreELEC (official): 8.90.015 (RK3399.arm)

    21:31:56 up 30 min, load average: 1.34, 1.57, 1.41

    78.125 Celsius

    My N2:

    CoreELEC (official): nightly_20190607 (Amlogic-ng.arm)

    21:41:50 up 29 min, load average: 2.28, 2.04, 1.73

    37.500 Celsius

    that's all I have to say ;)

  • Thank you all for replying!


    On additional note on the Khadas website, under the specifications as starred (*) additional information it says:

    [2] LPDDR4 or LPDDR4X RAM will be selected randomly during manufacturing.

    I know stuff are randomized in production all the time, but I find it a bit weird they go upon the RAM on which the LPDDR4X one is much more efficient if you want some kind of portability with a battery and also because I think LPDDR4X is not backwards compatible meaning that the boarding having this one will be different then the one that can have LPDDR4, am I missing something?

  • HK has been around longer and the likelyhood of diminishing software support is more of a worry with Khadas's products then a company like HK thats already established itself.

    I think Khadas is concentrating to much on the extraneous extra's market when really the bulk of their time should be spend on building and attracting better Software support at this point in the game. Ya i know HK sells extra's as well but they have been around longer and are much more settled in at the production level and development...

    Don't get me wrong as i am not saying Khada's products are not good its just they seem more focused on leveraging the various revenue streams leaving Core Software development to a very small group of coders. and based on the costs there is no way i would spend more on a Khadas board over a N2 at this point in time.

    I think currently Khada's position in the market really is just being pushed by a couple of individual's that for whatever reason seem to think they are superior to others when the reality is more based in who's getting the samples and whats expect of them. maybe down the road when Khada's support grows to include more then just a small group of dudes trying to support it that things will be better.

    To be honest if i was really looking for a S922 based box with anything to do with Dvb Streaming then i would be looking at the Dreambox One as its got the horsepower and resources to hardware handle a dvb transport stream properly and do all the processing that needs to be done.

    I am at this point tho unaware of anyone thats released LibreElec for it. tho i do know its been hacked to run Kodi.

  • Thank you all for replying!


    On additional note on the Khadas website, under the specifications as starred (*) additional information it says:

    [2] LPDDR4 or LPDDR4X RAM will be selected randomly during manufacturing.

    I know stuff are randomized in production all the time, but I find it a bit weird they go upon the RAM on which the LPDDR4X one is much more efficient if you want some kind of portability with a battery and also because I think LPDDR4X is not backwards compatible meaning that the boarding having this one will be different then the one that can have LPDDR4, am I missing something?

    Ya i would agree that seems a bit weird to make a disclaimer like.

    without comparing the actual specs between the 2 to see the differences its hard to say but typically whenever i have seen disclaimers like that it usually equates to parts-supply chain stuff where depending on the availability of the part at that point in the production cycle maybe what they are talking about. that would be my quess.

    • Official Post

    If you can't read on CE forum what is the reason to not support this device (yet) with "old" kernel it is your fault. If you don't understand ask someone to translate the reason to your language.

    You again give inaccurate information (you have previously tried to intentionally spread fakes about N2, after that I am skeptical about your information). As far as I know, the representative of Beelinc offered CE sample KING, but CE representatives said they need a lot of samples for all developers. This requirement is very similar to the attempt to put obviously impossible requirement that would then hide behind this, why they were not able to release the CE image for King.


    I must correct you (as others did many times) ... I'm no fanboy but rather manufacturer & SOC agnostic; always looking for the hardware which suits my needs best. So: just an ordinary user and I still have to say that the N2 cooling / temperature curve is superior to anything else I saw ...


    Comparison; both more or less in some sort of idle mode:

    I have explained many times what "not an efficient cooling system"means. I will repeat for you an example. Maybe then you'll understand what I'm writing. You bought a car. He has a gas tank the size of 100 liters. You filled 100 liters of gasoline in this car. We drove the distance and spent 40 liters. The car stopped going because it ran out of gas, although the tank has another 60 liters. Due to the poor design of the fuel tank, the car can not use the remaining 60 liters. The car can only use 40 liters. You will always have to fill it with 40 liters, the remaining 60 you will carry as ballast. The situation is similar with the cooling system N2. when used in normal mode (when the radiator is looking down), its efficiency is about 40%. It works better if put vertically (using a VESA mount). I ran tests. If you replace the regular radiator with another, which is less than the standard 2 times in area and less than 3 times in massa and place it correctly (up the radiator), when running the tests, I get even a little less temperature than when using the standard radiator. By the way, reducing the impact of high temperature on other components also has a positive effect on the overall performance. Compilation of the project (which lasts an average of 7-8 hours), with the redesign of the cooling system is reduced by about 10-15 minutes. This is not much, but it confirms that the decrease in temperature of all components has a beneficial effect on the system.

    p.s.

    I'm tired of explaining the truisms of the cooling systems. In the future, I see no reason to enter into a discussion if people do not know the basic laws of physics. :)

  • You again give inaccurate information (you have previously tried to intentionally spread fakes about N2, after that I am skeptical about your information). As far as I know, the representative of Beelinc offered CE sample KING, but CE representatives said they need a lot of samples for all developers. This requirement is very similar to the attempt to put obviously impossible requirement that would then hide behind this, why they were not able to release the CE image for King.

    ..............when used in normal mode (when the radiator is looking down), its efficiency is about 40%.

    "intentionally spread fakes": Where and what exactly? I only pointed to your (wrong) speculations.

    Company makes big profit from other people's work and they can't send few samples? Only one! If they are serious they should send few samples. And not "lot of samples" :)

    Update: It is alive 8|

    External Content youtu.be
    Content embedded from external sources will not be displayed without your consent.
    Through the activation of external content, you agree that personal data may be transferred to third party platforms. We have provided more information on this in our privacy policy.

    How did you calculated or measure exactly 40%? So others can repeat the tests.

  • Quote

    You again give inaccurate information (you have previously tried to intentionally spread fakes about N2, after that I am skeptical about your information). As far as I know, the representative of Beelinc offered CE sample KING, but CE representatives said they need a lot of samples for all developers. This requirement is very similar to the attempt to put obviously impossible requirement that would then hide behind this, why they were not able to release the CE image for King.

    Please refrain from speculating about something you obviously know nothing about. Your conclusion is grossly inaccurate.

    As to you comment on the N2 cooling, you can put forward any amount of theories as to why it wont work - but the real world empirical evidence says it does - and this has been demonstrated to you on numerous occasions. If you want to drive it hard as you suggest - turn it over mount a 12V 4inch fan run at 5V (a very simply DIY job) and you will achieve a steady below 50 Cent temps under any load.
    Meanwhile the VIM3 which you are advocating as a better product will sit at over 60 Centigrade in normal use and much higher in active use simply because it has NO COOLING. The VIM3 in stock form is on a fast track to burnup - now that is bad design.

    Shoog

  • "intentionally spread fakes": Where and what exactly? I only pointed to your (wrong) speculations.

    Some of the users (who are not familiar with the communication style of CE apologists) may have believed it .... :)

    Trash-talk

    Company makes big profit from other people's work and they can't send few samples? Only one! If they are serious they should send few samples. And not "lot of samples"

    Rhetorical question. If you have a working system for a similar model (N2), adding support for King is such a difficult task that it requires the creation of the whole system and the kernel again (requires a lot of work of all CE specialists) ?


    How did you calculated or measure exactly 40%? So others can repeat the tests.

    Please refrain from speculating about something you obviously know nothing about. Your conclusion is grossly inaccurate.


    As to you comment on the N2 cooling, you can put forward any amount of theories as to why it wont work - but the real world empirical evidence says it does - and this has been demonstrated to you on numerous occasions. If you want to drive it hard as you suggest - turn it over mount a 12V 4inch fan run at 5V (a very simply DIY job) and you will achieve a steady below 50 Cent temps under any load.
    Meanwhile the VIM3 which you are advocating as a better product will sit at over 60 Centigrade in normal use and much higher in active use simply because it has NO COOLING. The VIM3 in stock form is on a fast track to burnup - now that is bad design.

    If you are not ready to read the essence of what is written, it is not my problem. :)

    I don't see the point in wasting my time on idle chatter with the deaf , I'd rather spend that time doing useful work. For reference, I have a real sample of N2, which I intensively tested in different modes. Together with the HK developers, I did a useful job of improving the N2's user experience (developing and testing direct booting systems from USB, which should soon be published for General use). And I continue to develop Armbian and LE for N2.

    p.s. I'm not interested in communicating with you. :)

  • When samples are received of the King box (as they will be) then full support will be in CE in a matter of weeks - as it was with the N2. That is where you are spreading more misinformation beyond your knowledge of reality.

    No one here has questioned your efforts to further development of the mainline S922 kernel, but simply stop spreading bullshit about a cooling system which performs better than any other of the available boards. Your comments regarding the heatsink are simply laughable.

    You only seem interested in spreading misinformation and puffing your own work up.

    The only reality that should matter in making a choice of developer board here is that the N2 is ready to run as a stable, reliable, high performance media player straight out of the box.

    Shoog

    Edited 2 times, last by Shoog (June 11, 2019 at 12:33 PM).

  • If the boxes are working fine and not burning up i fail to see the reason to keep confusing people with a debate that is really based on nothing more then Opinions. and we all know that opinions are like assholes and we all have one..

    There are many ways to thermally cool something and who's to say which way is right, as long as the method chosen is capable of keeping things within the hardware's operating conditions it really makes no difference.

    The car analogy I must have lost something in the translation because I fail to see the point other then i quess your saying things can go wrong i quess.

    The manufactures make these boxes for pennies on the dollar in quantity and its not going to kill them to write-off a bunch of samples to developers as its typically done with other products. Or if they want to cheap out then maybe they should actually hire a bunch REAL coders and stop trying to suck off the public tit hoping to get the popular software ported to their products.

    As much as i believe in Open Source this is the part that makes people close up and keep stuff private as this whole particular industry is driven by basically Hardware Manufactures that really only provide rudimentary basic software and leave most everything else up to the Open Source Community while they create their next great device.

    Anyways the thermal debate is going nowhere as its become more of a who's got the biggest dick argument and its confusing people with speculation really based on a Opinion

    Edited once, last by buzzmarshall: change public to private on my open source statement (June 11, 2019 at 3:43 PM).

  • 100% Dead On.

    Normally I don't like saying things that may have a impact in what others purchase as i personally believe people should be left to choose for themselves based on what Their needs are.

    But i will say as someone that gets paid to evaluate hardware for others looking to develop devices and as the owner of a full fab rework lab that the quality of the HK stuff is superior to Khadas at this Point in time in my opinion. Now thats not to say that Khadas over time can't achieve the same level of quality as their engineering gets more mature.

    Balbes. as much as myself and others respect your contributions to the scene I do realize you've been comp'd by Khadas and are one of their resident guru's, but it seems a bit irresponsible on your part to use it as a platform to push one product over the other over something as debatable as the method of thermally cooling a device.

    Its like trying to argue over who's better LibreELEC or CoreELEC which are really just forks of the original OpenELEC. to me neither is better as it depends on what you want to do and the type of device your using because they both are currently on different adgenda's which is fine. to me CoreELEC seems to have settled in on getting the best working solutions at this point in time while LibreElec is concentrating on trying to unify things with mainline. To me that implies CE is more about people and working devices in the now whereas LE is willing to hold back and wait while unification to mainline is being worked on. Personally i think people are perfectly satisfied with boxes that run the newest kodi versions and are pretty much oblivious to the fact that its Not running on mainstream linux as that whole idea is really about making it easier to support more devices down the road and change the development cycle. but i do Get what LE is trying to do and understand why.