October 17, 2019, 11:23:49 am

News:

Have you visited the Allwinner Chipset wiki? - http://linux-sunxi.org/


Show posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - ssvb

1
Quote from: ua3prq on February 05, 2015, 03:05:14 am
Quote from: ssvb on February 03, 2015, 07:56:13 pm
Ну или как совсем простой варинт, попробуйте образ SD карты из https://github.com/ssvb/sunxi-bootsetup/releases/tag/20141215-sunxi-bootsetup-prototype


Спасибо за разработку!
Видимо какой-то новый дистрибутив намечается, да еще с инсталлятором ? ;-)

Тут скорее не дистрибутив, а аналог BIOS SETUP из десктопных компьютеров. С тестами для диагностики исправности железа. Который можно будет прикрутить к любому дистрибутиву или инсталлятору. Я туда в нагрузку только что добавил стандартный debian installer :)

QuoteТеперь о результатах на исправном железе:

CB3 A20.B- не грузится собственно юбут (UART log):
U-Boot SPL 2014.10-01706-gf75b83b (Jan 26 2015 - 10:58:21)
DRAM: 2048 MiB
CPU: 384000000Hz, AXI/AHB/APB: 1/2/2
я
U-Boot SPL 2014.10-01706-gf75b83b (Jan 26 2015 - 10:58:21)
DRAM: 2048 MiB
CPU: 384000000Hz, AXI/AHB/APB: 1/2/2

Спасибо! Оказывается, что Cubietruck не держит сильно низкую частоту памяти. Поэтому её пришлось поднять с якобы безопасных 312MHz до 360MHz.

QuoteТрак в дальнейшем доступен для экспериментов, будут новые мысли- пишите, попробуем. CB2 завтра отправляю снова на сервер, он постоянно занят.
Удачи!

Сейчас эта проблема уже должна быть исправлена.
2
Советую попробовать mainline u-boot. Не факт, что поможет (если железо действительно неисправно), но всё же есть какой-то шанс. По сравнению с u-boot-sunxi, там исправлены некоторые ошибки в коде (вроде некорректных задержек при ресете и прочих безобразий): http://lists.denx.de/pipermail/u-boot/2014-July/183981.html


git clone git://git.denx.de/u-boot.git
cd u-boot
# compile
make CROSS_COMPILE=arm-linux-gnu- Cubietruck_defconfig
make -j4 CROSS_COMPILE=arm-linux-gnu-
# write to SD card
sudo dd if=u-boot-sunxi-with-spl.bin of=/dev/sdX bs=1024 seek=8
sync


Ну или как совсем простой варинт, попробуйте образ SD карты из https://github.com/ssvb/sunxi-bootsetup/releases/tag/20141215-sunxi-bootsetup-prototype

Если эта SD карта загрузится, то в заголовке окна на HDMI мониторе будут отображены параметры памяти. Что-то вроде http://linux-sunxi.org/images/e/e3/20141215-sunxi-bootsetup-prototype-hdmi-menu.jpg
3
Quote from: null on January 16, 2015, 03:44:26 pm
dram_bus_width = 32
dram_io_width    = dram_bus_width / (кол-во чипов памяти)

в cubieboard - 2-а чипа => dram_io_width = 16
в cubietruck  - 4-е чипа => dram_io_width = 8

Всё верно. В cubietruck стоят 4-е чипа DDR3 x8 (восемь линий на шине данных) и плотностью 4096 каждый. В cubieboard стоят 2-а чипа DDR3 x16 (шестнадцать линий на шине данных) и плотностью 4096 каждый.

А теперь фокус-покус. Склейка из двух чипов DDR3 x8 является эквивалентом одного чипа DDR3 x16 вдвое большей плотности и их невозможно различить программно. Поэтому для cubietruck настройки "dram_io_width = 8 / dram_density = 4096" и "dram_io_width = 16 / dram_density = 8192" работают одинаково.
4
Часть параметров памяти автоопределяется на старте и успех инициализации может в некоторой степени зависеть от температуры чипа: http://linux-sunxi.org/A10_DRAM_Controller_Calibration (ключевое слово 'temperature').

Однако такие фатальные проблемы с загрузкой являются ненормальной ситуацией. Что-то там явно нестабильно.
5
Quote from: bug on September 19, 2014, 06:43:50 pm
Well, maybe different people can have a different problem and the cpufreq trick didn't work for 'em?

Sloop means that the other people have solved this problem ages ago and you are now re-discovering exactly the same solution :)
6
Quote from: foef on September 15, 2014, 12:14:18 am
The cubietruck boots and does everthing, just hangs from time to time (thats the nasty kind of fault as it cannot be debugged well).

The following factors can be eliminated (as the problem is fixed by swapping the cubietruck board)
- Power supply
- Network cable/connection
- Harddisk
- Operation system

These factors can't be eliminated yet, because different components may have slightly different tolerances. For example, if your PSU is bad, then one board may manage to cope with it under relatively light load, but another may fail in a much more obvious way.

Anyone who is suspecting hardware problems, may want to give http://linux-sunxi.org/Hardware_Reliability_Tests a try.
7
Quote from: seppel on September 15, 2014, 02:59:47 pm
i experience the same problems.

as soon as i try to use gigabit lan, i have a lot of network stability issues.

Are you by chance overclocking the CPU? http://www.cubieforums.com/index.php/topic,2590.0.html
8
Начинающим / Re: Я разочарован.
September 15, 2014, 02:56:36 pm
Quote from: luminofor on September 15, 2014, 09:42:12 am
Кстати про результат. Там про какой то Qt5 Cinematic Experience. А что это такое вообще? Первый раз слышу  :o

Qt5 (фреймворк для разработки приложений с пользовательским интерфейсом) поддерживает аппаратное ускорение с использованием OpenGL ES.

Qt5 Cinematic Experience - это всего лишь одно из небольших демонстрационных приложений для Qt5. Например, в X11 десктопе на Allwinner A10 оно выглядит так - https://www.youtube.com/watch?v=4SKrv2sl47I
Если совсем не нужны окна, то Qt5 работает и в чистом фреймбуфере. А также через libhybris (заднепроходный метод загрузки андроидных драйверов) оно должно прикручиваться и к Wayland.

Теперь осталось дело за самой малостью. Кто-то должен со времением переписать все линукс десктоп приложения на Qt5. И тогда будет всем нам аппаратно ускоренное графическое счастье :) Однако этот мифический "кто-то" пока валяет дурака. И я до сих пор не видел каких-либо реально полезных open source Qt5 приложений :(

Инструкции по сборке Q5 (в том числе для Cubieboard/Cubietruck) - http://linux-sunxi.org/Qt5_For_Mali_Binaries
9
General Discussion / Re: A10's/A20's cedarX etc. dead?
September 14, 2014, 07:08:56 pm
Quote from: murderered on September 14, 2014, 02:28:27 pm
mmmhhh... Now everything gets a bit clearer. Do you mean CedarX is "already sufficiently reverse engineered"? Completely?

http://linux-sunxi.org/Cedrus#Supported_codec_matrix

H264 is the most important and resource demanding codec. MPEG4/DIVX/XVID is lightweight enough to be successfully decoded even in software without hardware acceleration. And video overlays (disp layers) are used for offloading hardware scaling and colorspace conversion for any video codec (but this is not a part of cedar, but is provided by the display driver).

QuoteI'll try to set up an environment and trying out mpv. Maybe there is a solution for building a small media center on top off mpv.

There are some topics in this forums. And also some ready-made images, which can be used to demonstrate/test hardware accelerated video decoding :)
10
General Discussion / Re: A10's/A20's cedarX etc. dead?
September 14, 2014, 12:39:17 pm
Quote from: murderered on September 14, 2014, 10:51:33 am
Yesterday, I've read following in the xbmc-forum:
QuoteThe basic verdict is that no core XBMC dev has the desire to take up the task of porting the vidon.me code to XBMC.
A10/A20 being already quite outdated SoC's, it's just not worth it.

( http://forum.xbmc.org/showthread.php?tid=165232&pid=1719136#pid1719136 )

The XBMC guys are just holding a grudge against Allwinner:

"Now I can't speak for Gimli as I have not yet personally spoke with him as of yet, but for me it is over. I really can't stand lying people and therefor withdraw my hand of any cooperation with Allwinner.

There are already efforts of reverse engineer the VPU instructions. I will hope for that to materialise anytime soon. A second note; Don't get your hopes up for omenmax support from Allwinner anytime soon as the Xiaocuan asked for Gimliś help about openmax at the very first email contact already.

For me this about the last update about running a native linux/xbmc port on the Allwinner devices.

I am very sorry guys, but Allwinner just Killed this project."


And the XBMC guys are now venting their anger on the end users by deliberately boycotting Allwinner hardware even though it had been already sufficiently reverse engineered.

QuoteAfter searching a bit, there were nearly no new developments last months for cedarX, lbvdpau etc. needed for a open source video playback.

This last libvdpau-sunxi commit is only 12 days old: https://github.com/linux-sunxi/libvdpau-sunxi/commit/ea45c7887a4eeed111f88420d66a9d9e8b2429e3

As a matter of fact, libvdpau-sunxi works fine for the most important video formats since a long time ago. And you can use mplayer/mpv with SMPlayer GUI to get a nice hardware accelerated video playback :)

Simply forget about XBMC. That is unless you have the necessary skills and motivation to work on adding cedarx hardware acceleration support to it yourself.
11
You need to have a sunxi-3.4 kernel with the following commit: https://github.com/linux-sunxi/linux-sunxi/commit/9c77e95e1e60d13699a67e440327b0b664eb9ff6
As we can see at the github page, this patch is only included since "sunxi-v3.4.79-r1".
12
Quote from: rgmhtt on August 28, 2014, 11:20:24 am
After following the uboot list for a couple weeks, I am seeing strong involvement by freescale people, and can't tell what allwinner activity is going on.

Allwinner hardware is codenamed as 'sunxi', that's what you need to look for in the u-boot and linux kernel mailing lists. There are some pretty big sunxi improvements coming in u-boot 2014.10
13
Quote from: Sloop on July 06, 2014, 08:29:49 am
it would be nice if anyone could investigate some more tests while slightly raise the overclocking to acceptable values. So we know how much overclocking is allowed without leading the cubietruck to crash.

As mentioned earlier in this thread, pretty much every Allwinner A10 chip fails the cpufreq-ljt-stress-test at the clock speeds higher than 1.1GHz - https://www.mail-archive.com/[email]linux-sunxi@googlegroups.com[/email]/msg05843.html
Also confirmed by the results from lioka - http://irclog.whitequark.org/linux-sunxi/2014-07-03#9501794;

And this is only a single test program, which is not necessarily the most demanding. So having some CPU clock speed safety margin is a good idea. The default choice of 1GHz is very much reasonable. Allwinner A20 has even lower default clock speed than Allwinner A10. Neither of them can be clocked at 1.2GHz reliably.

The Chinese sellers from AliExpress did a lot of damage mis-advertising the unrealistic 1.2GHz and even 1.5GHz clock speeds for Allwinner devices. So people may feel encouraged to try these higher speeds and hurt themselves, just like we could see in this thread.
14
Quote from: slovenia on June 30, 2014, 09:39:37 am
I forgot the most obvious reason. :P SoC is slightly over-clocked.

1.2GHz is not a slight overclock. It is far beyond normal operational limits of A10/A20 and is almost guaranteed to fail on certain heavy workloads. You can try the tests from the following thread for the start: https://www.mail-archive.com/[email]linux-sunxi@googlegroups.com[/email]/msg04343.html

Yes, it is possible to improve reliability by increasing the core voltage, but then you have to deal with major overheating problems on heavy workloads (say hello to large heatsinks and active cooling if you want to resolve this): https://www.mail-archive.com/[email]linux-sunxi@googlegroups.com[/email]/msg01029.html

To sum it up, you get some really serious reliability problems for merely 20% of potential performance improvement.
15
Quote from: madmax on June 02, 2014, 03:52:38 pm
Does anybody know where to get the kernel sources (repo, changset, used config) for the kernel used in the following image?

http://dl.cubieboard.org/software/a20-cubietruck/lubuntu/ct-lubuntu-card0-v1.03/EN/

Using this image, Wlan on the CT works. With all kernels I built from several repos I wasn't able to get wlan to work.

I think that this is the right link: http://docs.cubieboard.org/tutorials/ct1/development/compiling_latest_kernel_for_cubietruck_cubieboard3

And recently some people expressed some interest in having Cubietruck WLAN supported in the the linux-sunxi kernels too: https://www.mail-archive.com/[email]linux-sunxi@googlegroups.com[/email]/msg05275.html
So, optimistically, this may be resolved soon and you will not have to hop some fancy git trees :)