UDOO Neo was unveiled last February as the first hobbyist board features Freecale i.MX6 Solox Cortex A9 + Cortex M4 processor. I was expecting UDOO to design support board since their earlier UDOO board combined Freescale i.MX6 processor with an Atmel MCU, and the new processor allowed for integrating the same functionality into a single chip. The board is now on Kickstarter where you can pick UDOO Neo Basic for $49 (Early bird is $35), or UDOO Neo for $59 (Early bird was $45) adding an Ethernet port, some sensors, and 1GB RAM, instead of 512 MB for the Basic version.
But both versions of the board basically share the same specifications:
SoC – Freescale i.MX 6SoloX ARM Cortex-A9 core @ 1GHz with 2D/3D GPU and ARM Cortex-M4 Core @ 166 MHz
Misc – Coin Cell RTC Battery Connector, Green Power Status LED, Configurable Red LED
Power Supply – 5V DC Micro USB; 12V DC power jack
Dimensions – 85mm x 59.3 mm
The board features Arduino compatible headers, and can be programmed with an Arduino IDE running on a separate PC or in the board itself. It has similar functionalities as the Raspberry Pi as it runs Linux (and Android), and offers similar interfaces, but adds Wi-Fi, Bluetooth Smart, and 9-axis motion sensors. So if you have a project that requires the power of Linux, and the I/O flexibility of Arduino, UDOO Neo boards should cost a little bit less than competing solutions, be easier to configure, and provide a more compact solution.
The ARM Cortex A9 core will run Android 4.4.3 + Linux 3.10, with UDOObuntu distribution to become available before the board ships, and the ARM Cortex M4 should run MXQ RTOS. Android and Linux source code will be provided. They also claim UDOO Neo will be open source hardware like the original UDOO. However, a Google search for the older board only shows UDOO schematics in PDF format, but after checking a bit more, I found the documentation page where the Gerber files, BoM, and mechanical files are also freely downloadable. Since the original schematics are not available, it’s not 100% open source hardware, but it’s still better than what is provided for the Raspberry Pi boards.
Since it’s UDOO project team have been around for a while, there’s already an active community, and several example projects for the older boards, but many should be adaptable to the Neo boards, and since it’s Arduino compatible, you can also leverage existing Arduino libraries and sketches.
The Kickstarter campaign started yesterday, and they already raised over $40,000 out of their $15,000 goals. Beside the pledges for UDOO Neo boards, they also have various kits including one with a 7″ LCD touchscreen display, a power supply, and cables, and bundles with up to 5 boards. Delivery is scheduled for September 2015.
Popcorn Hour VTEN a Linux based media player powered by Sigma Designs SMP8757 ARM Cortex A9 processor. Contrary to the many Android TV boxes I tested in the last few years, the device is dedicated to media playback, even though it has an App Store with 59 apps including YouTube, Facebook, and Twitter. I’ve already taken a few pictures of the device and board, so today, I’ll go through the user interface, quickly test the app store, go through my library of video test files, and check if HDMI audio pass-through is indeed working.
Popcorn Hour VTEN User Interface
I’ve connected VTEN to an Ethernet Gigabit switch, the HDMI port to Onkyo TX-NR636 AV receiver which itself is connected to LG 42UB820T 4K UHD TV, and the 5V/3A power supply. Pressing the On/Off switch at the back of the player will start it, and the boot takes just under one minute.
The user interface (1280×720 resolution) is quite simple, not necessarily a disadvantage for a media player, with the date, time and weather (if applicable) on the top, as well as icons on the bottom namely: Apps Market, Local Media, Recently Played, Shortcut, Setup, Setup Wizard, and Network Media.
I assume most people will simply ignore the Apps Market. It’s quite slow to navigate, and only features 59 apps to date, including YouTube, Facebook and Twitter app. However, if you don’t have another device to watch YouTube on your TV, you may still consider using it.
The Setup Wizard will allow you select your language, configure the network, and set the time and your location for weather. The Setup is fairly complete with preferences, personalization options (e.g. wallpaper, icon and text colors…), audio options with downmix / pass-through selection for most codecs including PCM, DTS and AC3 (Dolby), video options including resolution up to 1080p60 or 4K 30Hz, default zoom, aspect ratio, 3D output, and so on. You can also configure the network (IP / Proxy), manage network shares (SAMBA / NFS), configure parental control, set BD/Audio CD options, and upgrade the firmware.
The first time I booted the box, I quickly got a pop-up windows asking me if I wanted to upgrade the firmware. I went through the process, which downloaded the firmware and install it without any issues.
I used this “1 April 2015″ firmware for the complete review.
Local Media and Network Media allows you to respectively play files from USB or eSATA drive, or SAMBA or NFS shares. I did all testing from a SAMBA share, but while a USB flash drive was recognize, a Seagate USB hard drive with multiple partitions (NTFS / EXT-4 / exFAT / BTRFS) was simply ignore by the player. I could not test eSATA since I do not have any compatible drive. Finally the Shortcut is used for quick access to directories in Local Media or Network Media apps.
The best way to have a complete look at the user interface is to watch the video below, where I also test HDMI pass-through with DTS-HD MA 7.1 and Dolby TrueHD 7.1 videos, as well as 4K videos with H.264 and 10-bit HEVC / H.265.
Video Playback and Audio Pass-through Testing
I played my library of video test files from a SAMBA share running in an Ubuntu 14.04 computer with the Network Media app. The system had no issues finding my workgroups, and the PCs on the local network. The only inconvenience is that I had to input the username and password with the software keyboard using the remote control, but I guess that’s OK since it should be a one time thing.
I’ve started with some video samples from samplemedia.linaro.org, plus H.265 videos by Elecard, and a low resolution VP9 video:
MPEG2 codec / MPG container, 480p/720p/1080p – OK.
MPEG4 codec, AVI container 480p/720p/1080p – OK
VC1 codec (WMV), 480p/720p/1080p – OK
Real Media (RMVB), 720p / 5Mbps – RV8, RV9, and RV10 – Failed. Network Media app reports “No content found”. .rmvb files are probably filtered out
WebM / VP8 – Failed. Network Media app reports “No file listed”, as .webm files are filtered out
H.265 codec / MPEG TS container – 360p, 720p and 1080p – OK
WebM / VP9 (no audio in video) – Failed. Network Media app reports “No content found”. (webm file)
All files play very well, expect VP8/VP9 and RealMedia because the file extensions are supported, and did not show in the player.
Let’s move to some higher bitrate videos:
ED_HD.avi – Audio only.
big_buck_bunny_1080p_surround.avi (1080p H.264 – 12 Mbps) – OK.
h264_1080p_hp_4.1_40mbps_birds.mkv (40 Mbps) – OK
hddvd_demo_17.5Mbps_1080p_VC1.mkv (17.5Mbps) – Plays, but could be smoother (Most players are troubles playing this file smoothly).
Jellyfish-120-Mbps.mkv (120 Mbps video without audio) – OK
The Jellyfish video played fine over the Gigabit Ethernet connection, which is good news, however, I’ve seen some buffering with some other lower bitrate videos, so I wonder is Ethernet is 100% stable all of the time (TBC).
Most Android media player don’t support HD Audio pass-through via HDMI with codecs like DTS-HD or Dolby TrueHD, but Sigma Designs solution have been supporting HDMI audio pass-through for years. So I’ve check using Onkyo TX-NR636 AV receiver via HDMI and optical S/PDIF, and I’ve also made sure to check whether downmix is working for all tested codec.
Here are the settings I used for HDMI audio pass-through.
And the results of my tests:
Video’s Audio Codec
optical SPDIF Pass-through
AC3 / Dolby Digital 5.1
E-AC-3 / Dolby Digital+ 5.1
Dolby Digital+ 7.1
Audio Formats Not Supported over S/PDIF
Dolby TrueHD 5.1
Dolby TrueHD 7.1
DTS-HD High Resolution
So HDMI audio pass-through worked perfectly for me, and VTEN is the first device I’ve tested that support DTS-HD and TrueHD pass-through with 7.1 channels. However, if you don’t own an AV receiver, but still want to play video with Dolby TrueHD audio, you’re out of the luck, at least with this firmware version. The first line of the release notes of the firmware reads “Improve on Dolby TrueHD pass-through audio drop or audio-out-of-sync problems (reduce chance of happening)”, but I encountered this bug a few times during testing. The only solution is currently to stop the video, and resume where you stopped.
Sintel-Bluray.iso (unencrypted Bluray ISO) could play fine, as well as two 1080i MPEG2 videos (GridHD.mpg & Pastel1080i25HD.mpg). However, I could only get audio with Hi10p H.264 videos:
[Commie] Steins;Gate – NCED [BD 720p AAC] [10bit] [C706859E].mkv – Audio only / black screen. Shows 1,000 fps in Info overlay.
[1080p][16_REF_L5.1][mp3_2.0]Suzumiya Haruhi no Shoushitsu BD OP.mkv – Audio only / black screen. Lots of buffering despite low bitrate.
VTEN supports 4K output up to 2160p 30Hz via HDMI 1.4 output, so I tested a few 4K videos with relative success:
HD.Club-4K-Chimei-inn-60mbps.mp4 – OK
sintel-2010-4k.mkv – OK
Beauty_3840x2160_120fps_420_8bit_HEVC_MP4.mp4 (H.265) – OK (Appears cleaner than on other platforms, not white dots on black)
Bosphorus_3840x2160_120fps_420_8bit_HEVC_MP4.mp4 (H.265) – OK
Jockey_3840x2160_120fps_420_8bit_HEVC_TS.ts (H.265) – OK
MHD_2013_2160p_ShowReel_R_9000f_24fps_RMN_QP23_10b.mkv (10-bit HEVC) – OK, but seeking is not working.
phfx_4KHD_VP9TestFootage.webm (VP9) – Not listed in Network Media app
BT.2020.20140602.ts (Rec.2020 compliant video) – WBlack screen and no audio
tears_of_steel_4k_H264_24fps.mov – OK
big_buck_bunny_4k_H264_30fps.mp4 – OK
big_buck_bunny_4k_H264_60fps.mp4 – Plays OK (up to 30 fps), but I’ve noticed audio and video are out of sync
Overall 4K video playback is one of the best I’ve experienced so far. I’ve also check videos and pictures are not somehow downscaled but playing a 4K video test pattern, and taking a close up, as I did for 6 other ARM based 4K players.
The 4K video looks very sharp, and we can clearly see black and white pixels of the pattern video.
But jpg or png pictures are not quite as clear, so they must have been software decoded to the 720p frame buffer, instead of being rendered on an hardware scaler @ 4K resolution.
The latest Popcorn Hour model supports 3D videos, and despite my TV not supporting 3D, I’ve also checked 3D decoding capability of the platform:
bbb_sunflower_1080p_60fps_stereo_abl.mp4 (1080p Over/Under) – OK
bbb_sunflower_2160p_60fps_stereo_abl.mp4 (2160p Over/Under) – Black screen, audio only
Turbo_Film-DreamWorks_trailer_VO_3D.mp4 (1080p SBS) – OK
I’ve never manage to play the 2160p 3D video on any video, as it must require two 2160p hardware video decoders capable of decoding the top and bottom videos in the file.
Several AVI, MKV, IFO and MP4 videos (720p/1080p) could play smoothly, and without A/V sync issues. I had a bit less luck with FLV videos as only about 50% would work. One made the app restart, and I had a black screen and no audio for many of them. I also played a full 1080p movie (1h50 / MKV / 3GB) with any troubles. The box does not really get hot, and the maximum temperature I measured with an IR thermometer was 41 C.
You can checkout my “video samples” post for links to sample videos, especially the comments section.
Popcorn Hour VTEN is a pretty decent 4K media player, and it’s actually one of the best 4K and H.265 device I’ve tested so far, as well as the only one that supports both DTS-HD Master / High Resolution, and Dolby TrueHD HDMI audio pass-through. The player is also using the same high quality metal enclosure as in Open Hour Chameleon which ensure a relatively cool operation.
That’s not to say there aren’t any issues or limitations. For example, the device could not recognize any partitions on my hard drive (NTFS/ EXT-4 / exFAT / BTRFS), some video containers format are filtered out like webm and rmvb, and I had trouble playing several FLV videos, as well as Hi10p videos, and a black screen was all I got. I’ve also come across several bugs, such as losing audio after seeking or using trick mode while playing videos with HDMI audio pass-through enabled (Not reproducible with all files), and in one case I had video and audio were out of sync. Hopefully, Cloud Media will work through all these issues over time, and make VTEN an even better device.
Cloud Media (aka Syabas) sent me their latest Popcorn Hour VTEN media player. It supports HDMI output up to 4K, H.265 video codec, and features an eSATA connector, as well as optical and coaxial S/PDIF connectors. Today, I’ll show some pictures of the kit, and teardown the box to checkout the hardware, and in a few days, I’ll report on the user interface, video playback, and HDMI audio pass-through capabilities.
Popcorm Hour VTEN Unboxing
The unexpected device was sent via Fedex in the package shown below.
It lists some of the key features such as H.265, DSD (Direct Stream Digital), FLAC, Matroska, UPnP, 3D, and 4K Ultra HD support, as well as the specs, and more features and package content in English, French, Italian, Spanish, German, Dutch, Swedish, and Danish.
Popcorn Hour VTEN, Remote, Cables, Power Supply, User’s Guide, and Warranty (Click to Enlarge)
The box comes with an IR remote control and two AAA batteries, an HDMI cable, a SATA cable, a 5V/3A power supply, a quick start guide, and a warranty document.
Popcorn Hour V10 (Click to Enlarge)
The device itself looks very similar to Open Hour Chameleon, as they basically used the same metallic enclosure.
VTEN (aka V10) media player comes with two LEDs and the IR receiver window on the front panel, an SD card slot and a USB 2.0 host port on one side, with the rest of the port all available on the rear panel: an IR extension jack, optical and coaxial S/PDIF output, HDMI output, an Ethernet port, another USB 2.0 port, the eSATA port, an On/Off switch, and the power barrel.
Popcorn Hour VTEN Teardown
Opening the case is pretty easy, as all you need to do is to loosen four screws on the bottom.
Bottom of PCH V10 Board (Click to Enlarge)
The only interesting bit on this side is the MAC address with 00:06:DC prefix looking-up to “Syabas Technology (Amquest)”. To completely remove the board from the enclosure, you need to loosen two screws on the rear panel, and the four screws holding the PCB in place, before sliding the board out.
Just like with the Chameleon. thermal design has been neatly done with an aluminum ingot screwed to the metallic enclosure, and a thermal pad providing contact with the Sigma Designs processor.
Top of PCH V10 Board (Click to Enlarge)
The full name of the processor is “Sigma Designs SMP8757B80-CBE3 Secure Media Processor”. An SK Hynix H27U4G8F2ETR NAND flash provides 512MB internal storage, and two SK Hynix H5TQ4G63AFR DDR3 chips 1GB system memory. One of the two headers might provide access to the serial console for some hacking. There’s no wireless module for WI-Fi or Bluetooth, so an external USB dongle is required if you need wireless connectivity.
RaidSonic is a German company releasing products such as media players and multimedia accessories under their ICY BOX brand. One of those products is ICY Box IB-MP401Air music streaming received based on Atheros AR9331, and that looks very similar to SoundMate M2 I reviewed last year.
SoundMate M2 (Click to Enlarge)
But if you look on their product page, you’ll find out a few download links:
So I’ve downloaded the source code file (IB-MP401Air_Sources_and%20License_Terms.rar) to have a look. It has two compressed files, one with the license, and IB-MP401Air_Sources.tar.gz with the source code.
tar xzvf IB-MP401Air_Sources.tar.gz
-rwxr--r-- 1 jaufranc jaufranc 1400 Apr 16 16:49 building-the-firmware.txt
drwxr-xr-x 14 jaufranc jaufranc 4096 Feb 12 10:54 dns320B_GPL20150212
-rwxr--r-- 1 jaufranc jaufranc 925 Feb 12 08:25 making-u-boot.txt
The file showing how to build the firmware (OpenWRT) explains you should not use a newer Linux distribution:
Please note that this is an older OpenWrt version that might not build on newer distributions like Fedora 19 or later. If you want to build the firmware on a newer distribution you might need to get extra patches from OpenWrt to work around compilation errors. As an alternative you can install an older Linux distribution to build this firmware on.
Fedora 19 was released in July 2013, but I’ve still tried to build it on an Ubuntu 14.04 machine, and it failed:
x86_64-linux-gnu-gcc -std=gnu99 -I. -O2 -I/media/hdd/edev/sandbox/soundmate/dns320B_GPL20150212/dns320B_GPL20150212/staging_dir/host/include -O2 -I/media/hdd/edev/sandbox/soundmate/dns320B_GPL20150212/dns320B_GPL20150212/staging_dir/host/include -MT clean-temp.o -MD -MP -MF .deps/clean-temp.Tpo -c -o clean-temp.o clean-temp.c
In file included from clean-temp.h:22:0,
./stdio.h:477:1: error: ‘gets’ undeclared here (not in a function)
_GL_WARN_ON_USE (gets, “gets is a security hole – use fgets instead”);
It might build better on older Linux distributions, although you may still have to be ready to fix build errors on the way.
Yesterday, as I wrote a Linux 4.0 Changelog, I noticed a new ARM processor called AlphaScale ASM9260, and I thought it might be interesting to find out more. A Google search would only return results from the Linux kernel mailing list and mirrors, so I checked out the code a bit to find out its an ARM9 processor made by AlphaScale Integrated Circuits Systems, Inc, based in Wuxi, China. The company website alphascale.com does not work, but I found out the processor is sold on Taobao for 38 RMB (~$6), and there’s also a development board (ASM9260DVK) for 498 RMB or about $80, and that’s basically the only two links were I could find any details.
AlphaScale ASM9260T is arm ARM926EJS processor clocked at 240 MHz, 16GN I-cache and D-cache, 8KB SRAM, with the following peripherals:
The development board comes with a 10/100MEthernet interface (KSZ8051RNL EthernetPHY), micro and full size USB ports, two UART ports (DB9) for debugging and app?, a CAN port, a camera interface and two audio jacks (WM8731L Audio Codec). It can also support 4.5″, 7″ or 8″ displays, and boot from SPI NORandNAND. There’s also a 20-pin JTAG interface., and a few unpopulated UART and RS422 headers.
The software development kit description refers to ALPOS ADS1.2 development environment, but I can’t find any details about it, as well as support for Linux 2.6.38, U-boot and GCC cross-toolchain. It might not be exactly using the latest version of Linux, but since the developer who submitted patchsets for ASM9260 has been working on it since August 2014, some company must be serious about improving Linux support for the platform.
The company claims ASM9260 offers a more cost effective solution compared to Samsung S3C2440 and Atmel SAM9G45, partially because it can be used on 2-layer boards, while the competitors require 6-layer boards.
Apart from the two Taobao links listed in introduction, I can’t find any useful links, so in case you are interested in such platform, and need more details, you’d either have to call to +86(0)25 86892412/86462412 (in China) or QQ to 714960297, and you’d probably better speak Mandarin. [Update: There’s a product page in Chinese on Tianjili website].
We’ve already have two 96Boards compliant boards formally announced with HiSilicon Hikey and Qualcomm Dragonboard 410c, and Actions Semiconductor was also expected to release theirs soon. Albeit it’s not been officially announced yet, the company has been showcasing their Bubblegum-96 board at the Hong Kong Electronics Fair 2015.
Power Supply – ATC2609 PMIC; Should be 8-18V / 2A input as per 96Boards specs but I can’t see any power barrels.
Dimensions – 85 x 54 mm (96Boards compliant)
The board can run Android 5.0 or Ubuntu based on Linaro codebase. They did not provide the exact RAM and storage capacity, but since the board needs to be low cost, I’d use they’ll go with 1GB RAM, and 4GB internal storage, just like the Hikey board. There will also be another board called Actduino S900 powered by the same processor, running the same images, but not compliant with 96Boards specifications, and adding an Ethernet port, an LCD connector and so on.
So I decided to release 4.0 as per the normal schedule, because there really weren’t any known issues, and while I’ll be traveling during the end of the upcoming week due to a college visit, I’m hoping that won’t affect the merge window very much. We’ll see.
Linux 4.0 was a pretty small release both in linux-next and in final size, although obviously “small” is all relative. It’s still over 10k non-merge commits. But we’ve definitely had bigger releases (and judging by linux-next v4.1 is going to be one of the bigger ones).
Which is all good. It definitely matches the “v4.0 is supposed to be a_stable_ release”, and very much not about new experimental features etc. I’m personally so much happier with time-based releases than the bad old days when we had feature-based releases.
That said, there’s a few interesting numerological things going on with 4.0. Looking at just the statistics in git, this release is not just when we cross half a million commits total, but also cross the 4 million git object limit. Interestingly (if you look for numeric patterns), Linux 3.0 was when we crossed a quarter million commits and 2 million git objects, so there’s a nice (and completely unintentional) pattern there when it comes to the kernel git repository.
[ Another quick historical numerological footnote: the old historical BK tree was getting close to the 16-bit commilt limit that BK originally used to have. So that whole “quarter of a million commits” is actually quite a lot. During all of the BK years we only got 65k commits. Of course, we only used BK for three years, and we’ve now been on git for almost exactly ten years, but still – it shows how the whole development process has really sped up a _lot_ ]
Feature-wise, 4.0 doesn’t have all that much special. Much have been made of the new kernel patching infrastructure, but realistically, that not only wasn’t the reason for the version number change, we’ve had much bigger changes in other versions. So this is very much a “solid code progress” release.
Go get it and enjoy,
Linus “we’re all sheep” Torvalds
Linux 3.19 brought improvement to btrfs (raid), the network stack, added ARM Coresight, device tree overlays support, and more.
Some key changes made to Linux 4.0 include:
pNFS (Parallel NFS), UBIFS, F2FS and BTRFS File Systems improvements
Live Kernel Patching – Install kernel updates without rebooting
Intel Quark x86 SoC support
Various patches to improve Linux running on a Playstation 3
Open source AMD Radeon driver supports DisplayPort Audio and improves fan support
Some of the new features and improvements specific to the ARM architecture include:
HDMI output enabled on rk3288-firefly and rk3288-evb
Disable GMAC by default
Amlogic – pinctrl driver for Amlogic Meson SoCs
Regulator driver for Mediatek MT6397
Added watchdog driver
Added Mediatek MT8173 64-bit processor
New processors: Exynos 7, Freescale LS2085A, and Tegra 132 (Denver)
Various fixes for ARM64 including UEFI and KVM code.
Preparation work for Atmel AT91 support for multiplatform
Other new platforms – Alphascale ASM9260, Marvell Armada 388, CSR Atlas7, TI Davinci DM816x, Hisilicon HiP01, ST STiH418, and Conexant Digicolor (CX92755).
There has also been some interesting changes for the MIPS architecture:
Fixes for KVM support
Support for MIPS R6 processors
Preliminary support for Cavium Octeon 3 SoCs which feature up to 48 MIPS64 R3 cores with FPU and hardware virtualization
A more detailed changelog for Linux 4.0 will soon be available on Kernelnewbies.org, and once it’s up you may also want to have a look at their ARM architecture and drivers sections for more details about changes related to ARM, MIPS and other platforms. I’ve also generated a complete Linux 3.19 vs Linux 4.0 changelog (3.4MB) with git (comments only, no code).
I’ll just share some of their results, you can read the presentation, or go through the benchmark results to find out more.
HEP-SPEC06 Benchmark (Click to Enlarge)
HEP-SPEC06 is a new High Energy Physics (HEP) benchmark for measuring CPU performance developed by the HEPiX Benchmarking Working Group, and here it’s not surprising to see the low power solutions under-perform the more powerful Intel Xeon and Power 8 processors, with the latter taking the crown.
Geant 4 ParFullCMS (Click to Enlarge)
Geant 4 simulates the passage of particles through matter, something that you would expect the CERN to do regularly. Intel Xeon E3 outperforms IBM Power8 processor here.
But let’s move on to power consumption, and performance per watt.
Idle Power Consumption (Click to Enlarge)
IBM OpenPower 8 has a much higher power consumption than other systems, and HP Moonshot ARM 64-bit X-Gene 1 consumes more than both Intel servers. The chart under full load (not shown here) also shows a similar pattern.
HEP-SPEC06 per Watt (Click to Enlarge)
When it comes to performance per watt however, both HP Moonshot ARM and Power 8 systems are the least efficient here, and Intel systems provide the best ratio. Bear in mind that X-Gene 1 is manufactured with a 40nm process, while Applied Micro X-Gene 2 and 3 will be manufactured using 28nm and 16 nm FinFET processes, so some large efficiency gains could be expected here.
We may find out soon, as the CERN expects to add these two new processors, as well a Cavium ThunderX to their benchmarks in the future.