Quad Core i3-Based Type 6 COM Express Board

ADLINK has announced the addition of the quad-core Intel Core i3-8100H processor to its recently released Express-CF COM Express Basic size Type 6 module based on the 8th Generation Intel Core i5/i7 and Xeon processors (formerly Coffee Lake). The Express-CF/CFE is the first COM Express COM.0 R3.0 Basic Size Type 6 module supporting the Hexa-core (6 cores) 64-bit 8th Generation Intel Core and Xeon processor (codename “Coffeelake-H”) with Mobile Intel QM370, HM370, CM246 chipset.

Whereas previous generations Intel Core i3 processors supported only dual cores with 3 MB cache, the Intel Core i3-8100H is the first in its class to support 4 CPU cores with 6 MB of cache. This major upgrade results in a more than 80% performance boost in MIPS (million instructions per second), and an almost doubling of memory/caching bandwidth, all at no significant cost increase compared to earlier generations. Intel Core i3 processors are widely recognized as the best valued processor and are therefore preferred in high-volume, cost-sensitive applications. They are popular choices in gaming, medical and industrial control.

These Hexa-core processors support up to 12 threads (Intel Hyper-Threading Technology) as well as an impressive turbo boost of up to 4.4 GHz. These combined features make the Express- CF/CFE well suited to customers who need uncompromising system performance and responsiveness in a long product life solution. The Express-CF/CFE has up to three SODIMM sockets supporting up to 48 GB of DDR4 memory (two on top by default, one on bottom by build option) while still fully complying with PICMG COM.0 mechanical specifications. Modules equipped with the Xeon processor and CM246 Chipset support both ECC and non-ECC SODIMMs.

Integrated Intel Generation 9 Graphics includes features such as OpenGL 4.5, DirectX 12/11, OpenCL 2.1/2.0/1.2, Intel Clear Video HD Technology, Advanced Scheduler 2.0, 1.0, XPDM support, and DirectX Video Acceleration (DXVA) support for full H.265/HEVC 10-bit, MPEG2 hardware codec. In addition, High Dynamic Range is supported for enhanced picture color and quality and digital content protection has been upgraded to HDCP 2.2.

Graphics outputs include LVDS and three DDI ports supporting HDMI/DVI/DisplayPort and eDP/VGA as a build option. The Express-CF/CFE is specifically designed for customers with high-performance processing graphics requirements who want to outsource the custom core logic of their systems for reduced development time. In addition to the onboard integrated graphics, a multiplexed PCIe x16 graphics bus is available for discrete graphics expansion.

Input/output features include eight PCIe Gen3 lanes that can be used for NVMe SSD and Intel Optane memory, allowing applications access to the highest speed storage solutions and include a single onboard Gbit Ethernet port, USB 3.0 ports and USB 2.0 ports, and SATA 6 Gb/s ports. Support is provided for SMBus and I2C. The module is equipped with SPI AMI EFI BIOS with CMOS backup, supporting embedded features such as remote console, hardware monitor and watchdog timer.

ADLINK Technology | www.adlinktech.com

Signage-Oriented Mini-STX SBC Taps Ryzen V1000

By Eric Brown

Sapphire, which makes AMD-based graphics cards and motherboards, offers a 147.3 mm x 139.7 mm Mini-STX (5×5-inch) form factor SBC that runs Ubuntu 16.04 or Windows on AMD’s new Ryzen Embedded V1000 SoC. AMD’s Ryzen V1000 is highly competitive on CPU performance with the latest Intel Core chips, and the Radeon Vega graphics are superior, enabling four 4K displays to run at once.


FS-FP5V
(click image to enlarge)
The only other Ryzen V1000 based SBC we’ve seen is Seco’s open-spec, 120 x 120mm Udoo Bolt, which ships to Kickstarter backers in December. Sapphire’s commercial FS-FP5V is available for sale now with shipments beginning later this month, according to the Tom’s Hardware post that alerted us to the product.

The FS-FP5V starts at $325 for a model equipped with the dual-core, quad-thread V1202B version of the Ryzen V1000 with lower-end Vega 3 graphics. The three models with the quad-core, octa-threaded versions of the SoC go for $340, $390, and $450, with ascending clock rates and graphics ranging from Vega 8 to 11.



AMD Ryzen Embedded V1000 models, all of which are available with the FS-FP5V
(click image to enlarge)
Pricing, which does not include RAM or storage, seems to be a bit higher than the Udoo Bolt. The Bolt also adds an Atmega32U4 MCU for Arduino and Grove compatibility but is limited to the two lower-end V1000 SoC models. The Bolt seems more like a general purpose embedded board while the FS-FP5V, which has up to 4x DisplayPorts, is more directly aimed at digital signage and other media-centric applications including electronic gaming, medical imaging, thin clients, and POS terminals.

Unlike the Udoo Bolt, there’s no microSD slot or eMMC. There is however, a SATA III slot with power headers, as well as an M.2 M-key 2280 slot for SATA III or PCIe. A separate M.2 E-key 2242 connection supports PCIe devices including WiFi modules.

 
FS-FP5V portside views
(click images to enlarge)
It’s unclear if the cited prices include all four DP++ ports, which are listed as “up to 4x.” The board is further equipped with an audio jack, 2x GbE ports, serial and GPIO headers, and 3x USB 2.0 host ports. There’s also a USB 3.1 Type-C port, which does not appear to be used for DP. It’s unclear if it’s used for power.

 
Bleujour Kubb enclosure for FS-FP5V (left) and upcoming FS-FP5V-based 2×2 display wall from Seneca Data
(click images to enlarge)
This is Sapphire’s first Mini-STX SBC. Its other AMD-based motherboards include AMD R-Series based Mini-ITX boards and some 4×4-inch eNUC form factor boards such as the G-Series based LX 210.

In the YouTube video farther below, a Sapphire rep says his company can make custom boards based on the Ryzen V1000. The video also shows a Kubb enclosure for the FS-FP5V from Bleujour, as well as an upcoming 2×2 digital signage display wall from Seneca Data that taps the FS-FP5V to generate 4x 4K displays.

Specifications listed for the FS-FP5V include:

  • Processor — AMD Ryzen Embedded V1000 (see chart above)
  • Memory — 0GB to 32GB of dual-channel DDR4 RAM up to 3200MHz with ECC support via 2x sockets
  • Storage:
    • M.2 M-Key 2280 slot for SATA III or PCIe x4)
    • SATA III connector with 5V SATA power
  • Wireless — M.2 E-Key 2242 for WiFi and other PCIe x1
  • Networking — 2x Gigabit Ethernet ports (Realtek RTL8111G)
  • Display/media:
    • Up to 4x DisplayPort++ available via 2x dual-role USB 3.0 Type-C ports
    • 4x simultaneous 4K@60 displays
    • Radeon Vega 3, 8, or 11 graphics with DirectX 12, EGL 1.4, IOMMU 2.0, OpenCL 2.1, OpenGL ES 1.1, 2.x, and 3.x (Halti), OpenGL Next (Vulkan), OpenGL 4.6, 10-bit HEVC decoder (H.265), VP9 decoder, up to 10-bit, limited profile 2, Eyefinity
    • 3.5mm audio jack (ALC262 HD 4CH)
  • Other I/O:
    • USB 3.1 Type-C port
    • 3x USB 2.0 host ports
    • RS232/422/485 header
    • GPIO headers
  • Operating temperature — 0 to 50°C
  • Dimensions — 147.3 x 139.7mm
  • Operating system — Ubuntu 16.04 with Linux 4.9 or 4.14.14; Windows 7/8.1/10 etc.


 

AMD’s promo video for FS-FP5V
Further information

The FS-FP5V is available now starting at $325, with shipments due later this month. More information may be found at Sapphire’s FS-FP5V product page, which links to an order form.

This article originally appeared on LinuxGizmos.com on July 9.

Sapphire | www.sapphiretech.com

Variscite’s Latest DART Module Taps Headless i.MX6 ULZ

By Eric Brown

Variscite is spinning out yet another pin-compatible version of its 50 mm x 25 mm DART-6UL computer-on-module, this time loaded with NXP’s headless new i.MX6 ULZ variant of the single Cortex-A7 core i.MX6 UL. Due for a Q4 launch, the unnamed module lacks display or LAN support. It’s billed as “a native solution for headless Linux-based embedded products such as IoT devices and smart home sensors requiring low power, low size and rich connectivity options.”


DART-6UL with iMX6 ULZ 
(click image to enlarge)
The lack of display and LAN features mirrors the limitations of the i.MX6 ULZ, which NXP refers to as a “cost-effective Linux processor.” The headless, up to 900  MHz Cortex-A7 ULZ SoC offers most of the I/O of the of the i.MX6 UL/ULL, including ESAI, S/PDIF, and 3x I2S audio interfaces, but it lacks features such as the 2D Pixel acceleration engine and Ethernet controllers.


NXP i.MX6 ULZ block diagram
(click image to enlarge)
Last year, Variscite spun the Linux-ready DART-6UL into a faster, 696MHz v1.2 upgrade, which added the option of NXP’s power-efficient i.MX6 ULL SoC in addition to the i.MX6 UL. A few months later, it followed up with a DART-6UL-5G model that boasts an on-board, “certified” WiFi/Bluetooth module with dual-band, 2.4 GHz/ 5 GHz 802.11ac/a/b/g/n.


DART-6UL-5G (left) and DART-6UL v1.2
(click images to enlarge)
The upcoming i.MX6 ULZ based version, which we imagine Variscite will dub the DART-6ULZ, has the same Wi-Fi-ac module with Bluetooth 4.2 BLE. Like the latest versions of the other DART-6UL modules, the module can be clocked to 900 MHz.

The “cost effective” ULZ version differs in that it lacks the other models’ touch-enabled, 24-bit parallel RGB interface and dual 10/100 Ethernet controllers. Other subtracted features compared to earlier models include dual CAN, parallel camera, and “extra security features.”

The new module is also limited to a 0 to 85°C range instead of being available in 0 to 70°C or -40 to 85°C versions. The i.MX6 ULZ SoC itself has a slightly wider range of 0 to 95°C.

The pin compatible DART-6UL with iMX6 ULZ will offer the i.MX6 ULZ SoC with “optional security features,” which include TRNG, AES crypto engine, and secure boot. The 50 mm x 25mm module will ship with 512MB DDR3L, which was the previous maximum of the now up to 1 GB RAM DART-6UL. The storage range is similar, with a choice 512 MB NAND and up to 64 GB eMMC.

The DART-6UL with i.MX6 ULZ will support 2x USB 2.0 OTG host/device ports, audio in and out, and UART, I2C, SPI, PWM, and ADC interfaces. OS support is listed as “Linux Yocto, Linux Debian, Boot2QT.”

The ULZ version of the DART-6UL will support existing development kits, which are based on the VAR-6ULCustomBoard. This 100 mm x 70 mm x 20 mm carrier board offers a microSD slot, a USB host port, and micro-USB OTG and debug ports, as well as features that are inaccessible to the ULZ, including dual GbE, RGB, LVDS, CAN and camera interfaces.

This week Variscite announced another DART module based on another new NXP SoC. The DART-MX8M-Mini module taps a 14nm-fabricated i.MX8M Mini SoC variant of the i.MX8M with one to four 2GHz Cortex-A53 cores and a 400 MHz Cortex-M4, plus scaled down 1080p video via MIPI-DSI.

Further information

The DART-6UL with iMX6 ULZ will be available in the fourth quarter. The DART-6UL/ULL/ULZ product page notes that the lowest, volume-discounted price is $24, which likely pertains to the ULZ part. More information may be found in Variscite’s announcement.

This article originally appeared on LinuxGizmos.com on September 19.

Variscite | www.variscite.com

Rugged, Sandwich-Style SBC is Based on Sitara AM5718 MCU

By Eric Brown

Forlinx Embedded Technology, the Chinese company behind Linux-friendly SBCs such as the Texas Instruments (TI) Sitara AM3354 based OK335xS-II and the Forlinx i.MX6 SBC, has posted details on a new OK5718-C SBC. Like the OK335xS-II, it’s a Sitara based board, in this case tapping TI’s single-core, Cortex-A15 based Sitara AM5718. Like the i.MX6 SBC, it’s a sandwich-style offering, with the separately available FET5718-C module hosting the up to 1.5GHz AM5718.


OK5718-C
The OK5718-C was announced (translated) in China back in May, and the product page was recently spotted by CNXSoft. The FET5718-C module and OK5718-C SBC both support -40 to 85℃ temperatures and feature an optimized Linux distro with Linux 4.9.41, Qt 5.6, and Wayland. The BSP includes PCIe host and slave mode optimizations, a simplified file system for faster boot and flashing, and an image system to allow Weston virtual keyboards and easy Qt image stacking, says Forlinx.

FET5718-C module

The FET5718-C module’s Sitara AM5718 SoC may have a somewhat old-school CPU, but it provides plenty of extras. You get both a PowerVR SGX544 3D GPU and Vivante GC320 2D GPU, as well as a 750MHz TI DSP-C66X digital signal processor and video accelerator. There’s also the same, 200MHz programmable PRU subsystem found on the BeagleBone, as well as dual, 213MHz Cortex-M4 microcontrollers.



FET5718-C 

The combination of the DSP with the real-time MCUs enables robotics, machine vision, medical imaging, automotive, and facial recognition applications. Industrial automation and building automation applications are also supported.

The FET5718-C module adds 1GB DDR3L, 8GB eMMC, a TPS659162RGZR power management unit, and a 3-port Gigabit Ethernet switch subsystem. The 12-layer, 70 x 50mm COM runs on 5V power and has a 320-pin board-to-board connector.

OK5718-C board

The 4-layer, 190 x 130mm OK5718-C baseboard expands upon the FET5718-C features with ports popping out on all sides. The board provides 2x GbE ports, onboard WiFi and Bluetooth, and a mini-PCIe slot with optional 3G/4G. There are single USB 3.0 host and micro-USB 2.0 device ports and a pair of USB 2.0 host ports.

The OK5718-C is further equipped with an HDMI port, an SD slot, a CAN port, and dual audio jacks. Onboard I/O includes SATA 2.0 with power, DVP and 2x MIPI-CSI camera interfaces, and other I/O as detailed below.



OK5718-C detail view
(click image to enlarge)

Specifications listed for the OK5718-C SBC include:

  • Processor (via FET5718-C module) — TI Sitara AM5718 (1x Cortex-A15 core @ up to 1.5GHz; PowerVR SGX544 3D GPU; Vivante GC320 2D GPU; 750MHz TI DSP-C66X; IVA-HD image/video accelerator; 200MHz PRU-ICSS; 2x 213MHz Cortex-M4
  • Memory/storage:
    • 1GB DDR3L (via FET5718-C)
    • 8GB eMMC (via FET5718-C)
    • QSPI flash (via FET5718-C)
    • SD slot (SD, SDHC, SDXC support)
    • SDIO interface
    • SATA 2.0 interface with SATA power
  • Wireless — 802.11b/g/n with Bluetooth
  • Networking — 2x GbE ports
  • Media I/O:
    • HDMI 1.4a port for up to 1080P@60Hz
    • RGB 888 LCD interface
    • Dual display support
    • 2x MIPI-CSI
    • DVP 8-bit 5MP camera interface
    • Mic and headphone jacks; speaker headers
  • Other I/O:
    • USB 3.0 host port
    • 2x USB 2.0 host ports
    • Micro-USB 2.0 device port
    • 3x UART
    • 2x I2C
    • Serial debug port
    • CAN 2.0, SPI, GPMC, HDQ, JTAG
  • Expansion — Mini-PCIe slot with optional Huawei 3G/4G card
  • Other features — 2x LED; 3x user keys; RTC with coin-cell battery; boot config switch
  • Power — 12V DC input; power and reset switches
  • Operating temperature — -40 to 85°C
  • Dimensions — 190 x 130mm
  • Operating system — Custom Linux with Kernel 4.9.41, Qt 5.6, and Wayland

Further information

No pricing or availability information was provided for the OK5718-C SBC or FET5718-C module. More information may be found on the Forlinx OK5718-C and FET5718-C product pages. There’s also a product page at Faststream Technologies.

This article originally appeared on LinuxGizmos.com on August 20.

Texas Instruments | www.ti.com

Apollo Lake Pico-ITX SBC Sports Dual GbE Ports

AAEON has released the PICO-APL4, a compact SBC that features onboard memory and storage along with dual Gbit Ethernet support. The board is well suited for factory automation and IoT gateway systems. AAEON’s latest PICO form factor board is powered by an Intel Atom, Pentium N4200 or Celeron N3350 processor. By fitting the controller with up to 4 GB of onboard DDR3L memory and up to 64 GB of onboard eMMC storage, AAEON has made the board more resistant to the shocks and vibrations regularly experienced in industrial settings. This also cuts application development times because system developers don’t need to test the compatibility of external memory and storage.

The flexible PICO-APL4 houses USB ports and connectors, COM ports, a HDMI port and optional eDP. There’s also a 4-bit DIO to manage peripherals and an optional four-lane MIPI-CSI. Expansion is easily achieved with M.2 B and E keys enabling the connection of additional storage and WiFi/Bluetooth modules. Through an optional board-to-board interface, customers can also extend the IO interface and add a larger DIO.

Features:

  • Intel Atom/ Pentium N4200/ Celeron N3350 Processor SoC
  • Onboard DDR3L 2 GB (Optional to 4 GB)
  • Onboard Storage eMMC 16 GB (Optional to 32 GB / 64 GB)
  • Dual Gigabit Ethernet Support
  • HDMI 1.4b, eDP (Optional) for Display
  • BIO Reserved (Optional)
  • USB 3.0 x 2, USB 2.0 x 2, SATA 6.0 Gb/s x 1
  • 2 B Key (2280) x 1, M.2 E Key (2230) x 1

AAEON | www.aaeon.com

Industrial Mini-ITX Board Pumps up with Coffee Lake

By Eric Brown

Commell’s “LV-67X” Mini-ITX board runs on 8th Gen “Coffee Lake” processors, with up to 32GB DDR4, 3x SATA, triple 4K displays, USB 3.1, and PCIe x16 and mini-PCIe expansion. The LV-67X, which shares some of the layout and feature set of its Intel Apollo Lake based LV-67U board, is the first industrial Mini-ITX board we’ve seen with Intel’s 8th Gen Coffee Lake CPUs. (Going forward, we’ll likely use the caffeinated nickname rather than “8th Gen” because Intel also applies the 8th Gen tag to the transitional and similarly 14nm Kaby Lake-G chips as well as the new, 10nm Cannon Lake processors.)


LV-67X
(click image to enlarge)
The LV-67X is called an industrial board, and it provides a relatively wide 0 to 60°C range and a smattering of industrial I/O. However, it has a full-height profile and bridges the gap to consumer applications. The board supports video gaming, virtual reality, medical devices, imaging, machine vision, and digital signage. The product page lists only Windows drivers, but the manual notes that the board also supports Linux.

The 170 x 170mm SBC supports Coffee Lake Core, Celeron, and Pentium CPUs that work with the FCLGA1151 socket (the full name for LGA1151). The board ships with Intel Q370 chipset, one of Intel’s 300-series I/O chips announced with Coffee Lake that supports USB 3.1 Gen2 and extensive PCIe lanes.

No specific models were mentioned, but the SBC is said to support Coffee Lake chips with up to six cores running at up to 4.7GHz Turbo, with Intel 9th-gen graphics and up to 12MB cache. That would be the profile for the top-of-the-line Core i7-8700K, a hexa-core chip with 12 threads and a 95W TDP.

The LV-67X can load up to 32GB of speedy, 2666MHz DDR4 RAM via dual sockets. It provides 2x GbE ports, 3x SATA III interfaces, a full-size mini-PCIe slot with mSATA support, and another half-size mini-PCIe slot accompanied by a SIM card slot. There’s also a PCIe x16 interface.


 
LV-67X block diagram (left) and detail view
(click images to enlarge)

The description of the USB feature set varies depending on the citation, but Commell has clarified matters for us in an email. There are 6x USB 3.1 interfaces, 4x of which are coastline ports. There are also 4x USB 2.0 internal interfaces.

One key difference between earlier Core-based boards is that the LV-67X taps Coffee Lake’s ability to power three independent 4K displays. The board accomplishes this hat trick with coastline HDMI and DVI-I ports and an optional DisplayPort, as well as onboard VGA and 18/24-bit, dual-channel LVDS interfaces. If you don’t want the DisplayPort, you can instead get additional VGA and LVDS connections.

The LV-67X is further equipped with 4x RS232/422/485 or RS-232 interfaces, depending on conflicting citations, with an option to add two RS232/422/485 DB9 ports. Other features include 3x audio jacks (Realtek ALC262), 8-bit DIO, and LPC, SMBus, and PS/2 interfaces. You also get a watchdog, RTC with battery, and 24-pin ATX and 4-pin, 12V inputs.

Further information

No pricing or availability information was provided for the LV-67X. More information may be found on Commell’s announcement and product pages.

This article originally appeared on LinuxGizmos.com on August 17..

Commell | www.commell.com.tw

Raspberry Pi’s PoE HAT Ships for $20, Tosses in a Free Fan

By Eric Brown

Raspberry Pi Trading has launched a $20 Power-over-Ethernet HAT board for the Raspberry Pi 3 Model B+ that delivers up to 15W and ships with a small fan. The Power-over-Ethernet HAT that was promised with the release of the Raspberry Pi 3 Model B+ SBC has arrived. The $20, 802.3af-compliant “Raspberry Pi PoE HAT” allows delivery of up to 15W over the RPi 3 B+’s USB-based GbE port without reducing the port’s up to 300Mbps bandwidth.


 
Raspberry Pi PoE HAT alone (left) and fitted on Raspberry Pi 3 Model B+
(click images to enlarge)

We’ve seen an increase in the use of PoE in embedded equipment over the last year, perhaps due to the growth in IoT applications in which embedded gear must be placed in remote locations. It’s cheaper and easier to run Ethernet cable to a remote device than to extend electrical lines.

With the help of the RPi 3 B+’s improved PXE boot function, which enables network booting, “you can now dispense with not only the power supply but also the SD Card, making deployment even cheaper for a Raspberry Pi based system in your factory or workplace,” writes Roger Thornton in the Raspberry Pi blog announcement.

The Raspberry Pi PoE HAT features a fully isolated switched-mode power supply with 37-57V DC, Class 2 input and 5V/2.5A DC output. The HAT connects to both the 40-pin header and a new PoE-specific 4-pin header introduced with the B+ located near the USB ports. To enable PoE, you need power sourcing equipment, which is either “provided by your network switch or with power injectors on an Ethernet cable,” writes Thornton.


 
Raspberry Pi PoE HAT with RPi 3 B+ (left) and close-up of 4-pin header on B+ between the USB ports and the 40-pin GPIO header
(click images to enlarge)

The PoE HAT ships with a 25 x 25mm brushless fan for cooling the Broadcom SoC. This does not appear to be due to any additional heat generated by PoE. Instead: “We see the product as a useful component for people building systems that may be in tougher environments,” writes Thornton.

The fan is connected via I2C and controlled with an Atmel MCU chip. This setup turns on the fan automatically when the SBC hits a certain temperature threshold, a trick that requires the latest sudo rpi-update firmware.

You can add another HAT board on top of the fan with the help of some pass-through headers for the 40-pin GPIO and the 4-way header to expose the pins on the other side of the PoE HAT. Raspberry Pi Trading recommends the 2×20 pin header from Pimoroni and 4-way risers from RS and element14.

The Raspberry Pi 3 Model B+ won LinuxGizmos’ reader survey of 116 Linux/Android hacker boards. The community-backed SBC builds upon the RPi 3 Model B design with a faster, up to 1.4GHz quad-core Broadcom SoC, as well as faster Ethernet (GbE). You also get various power management improvements and faster dual-band 802.11ac and Bluetooth 4.2, which comes in a pre-certified, shielded module.

Further information

The Raspberry Pi PoE HAT is available now for $20 at a variety of resellers. The blog announcement may be found here. The product page with links to resellers is here.

This article originally appeared on LinuxGizmos.com on August 25.

Raspberry Pi Foundation | www.raspberrypi.org

Signage-Oriented Mini-STX SBC Taps Ryzen V1000

By Eric Brown

Sapphire has launched a Linux-friendly “FS-FP5V” SBC starting at $325 that features an AMD Ryzen V1000 SoC, as well as SATA III, 2x M.2, 4x DP++, 2x GbE, and 4x USB ports including a USB 3.1 Type-C.

Sapphire, which makes AMD-based graphics cards and motherboards, has launched a 147.3 mm x 139.7 mm Mini-STX (5×5-inch) form factor SBC that runs Ubuntu 16.04 or Windows on AMD’s new Ryzen Embedded V1000 SoC. AMD’s Ryzen V1000 is highly competitive on CPU performance with the latest Intel Core chips, and the Radeon Vega graphics are superior, enabling four 4K displays to run at once.


FS-FP5V
(click image to enlarge)

The only other Ryzen V1000 based SBC we’ve seen is Seco’s open-spec, 120 x 120mm Udoo Bolt, which ships to Kickstarter backers in December. Sapphire’s commercial FS-FP5V is available for sale now with shipments beginning later this month, according to the Tom’s Hardware post that alerted us to the product.

The FS-FP5V starts at $325 for a model equipped with the dual-core, quad-thread V1202B version of the Ryzen V1000 with lower-end Vega 3 graphics. The three models with the quad-core, octa-threaded versions of the SoC go for $340, $390 and $450, with ascending clock rates and graphics ranging from Vega 8 to 11.


AMD Ryzen Embedded V1000 models, all of which are available with the FS-FP5V
(click image to enlarge)
Pricing, which does not include RAM or storage, seems to be a bit higher than the Udoo Bolt. The Bolt also adds an Atmega32U4 MCU for Arduino and Grove compatibility but is limited to the two lower-end V1000 SoC models. The Bolt seems more like a general purpose embedded board while the FS-FP5V, which has up to 4x DisplayPorts, is more directly aimed at digital signage and other media-centric applications including electronic gaming, medical imaging, thin clients and POS terminals.

Unlike the Udoo Bolt, there’s no microSD slot or eMMC. There is however, a SATA III slot with power headers, as well as an M.2 M-key 2280 slot for SATA III or PCIe. A separate M.2 E-key 2242 connection supports PCIe devices including Wi-Fi modules.


 
FS-FP5V portside views
(click images to enlarge)
It’s unclear if the cited prices include all four DP++ ports, which are listed as “up to 4x.” The board is further equipped with an audio jack, 2x GbE ports, serial and GPIO headers, and 3x USB 2.0 host ports. There’s also a USB 3.1 Type-C port, which does not appear to be used for DP. It’s unclear if it’s used for power.


 
Bleujour Kubb enclosure for FS-FP5V (left) and upcoming FS-FP5V-based 2×2 display wall from Seneca Data
(click images to enlarge)
This is Sapphire’s first Mini-STX SBC. Its other AMD-based motherboards include AMD R-Series based Mini-ITX boards and some 4×4-inch eNUC form factor boards such as the G-Series based LX 210.

In the YouTube video farther below, a Sapphire rep says his company can make custom boards based on the Ryzen V1000. The video also shows a Kubb enclosure for the FS-FP5V from Bleujour, as well as an upcoming 2×2 digital signage display wall from Seneca Data that taps the FS-FP5V to generate 4x 4K displays.

Specifications listed for the FS-FP5V include:

  • Processor — AMD Ryzen Embedded V1000 (see chart above)
  • Memory — 0 GB to 32 GB of dual-channel DDR4 RAM up to 3200 MHz with ECC support via 2x sockets
  • Storage:
    • M.2 M-Key 2280 slot for SATA III or PCIe x4)
    • SATA III connector with 5 V SATA power
  • Wireless — M.2 E-Key 2242 for Wi-Fi and other PCIe x1
  • Networking — 2x Gbit Ethernet ports (Realtek RTL8111G)
  • Display/media:
    • Up to 4x DisplayPort++ available via 2x dual-role USB 3.0 Type-C ports
    • 4x simultaneous 4K@60 displays
    • Radeon Vega 3, 8, or 11 graphics with DirectX 12, EGL 1.4, IOMMU 2.0, OpenCL 2.1, OpenGL ES 1.1, 2.x, and 3.x (Halti), OpenGL Next (Vulkan), OpenGL 4.6, 10-bit HEVC decoder (H.265), VP9 decoder, up to 10-bit, limited profile 2, Eyefinity
    • 3.5mm audio jack (ALC262 HD 4CH)
  • Other I/O:
    • USB 3.1 Type-C port
    • 3x USB 2.0 host ports
    • RS232/422/485 header
    • GPIO headers
  • Operating temperature — 0 to 50°C
  • Dimensions — 147.3 x 139.7mm
  • Operating system — Ubuntu 16.04 with Linux 4.9 or 4.14.14; Windows 7/8.1/10 etc.


AMD’s promo video for FS-FP5V

Further information

The FS-FP5V is available now starting at $325, with shipments due later this month. More information may be found at Sapphire’s FS-FP5V product page, which links to an order form.

This article originally appeared on LinuxGizmos.com on July 9.

Sapphire | www.sapphiretech.com

The Voting Results are in. We Have a Winner!

Circuit Cellar’s sister website LinuxGizmos.com has completed its 2018 hacker board survey, which ran on SurveyMonkey in partnership with Linux.com. Survey participants chose the new Raspberry Pi 3 Model B+, as the favorite board from among 116 community-backed SBCs that run Linux or Android and sell for under $200.
All 116 SBCs are summarized in LinuxGizmos’ recently updated hacker board catalog and feature comparison spreadsheet.

GO HERE TO READ THE SURVEY RESULTS WITH ANALYSIS

Deadline Extended to June 22 — Vote Now!

UPDATE: We’ve extended our 2018 reader survey on open-spec Linux/Android hacker boards through this Friday, June 22.   Vote now!

Circuit Cellar’s sister website LinuxGizmos.com has launched its fourth annual reader survey of open-spec, Linux- or Android-ready single board computers priced under $200. In coordination with Linux.com, LinuxGizmos has identified 116 SBCs that fit its requirements, up from 98 boards in its June 2017 survey.

Vote for your favorites from LG’s freshly updated catalog of 116 sub-$200, hacker-friendly SBCs that run Linux or Android, and you could win one of 15 prizes.

Check out LinuxGizmos’ freshly updated summaries of 116 SBCs, as well as its spreadsheet that compares key features of all the boards.

Explore this great collection of Linux SBC information. To find out how to participate in the survey–and be entered to win a free board–click here:

GO HERE TO TAKE THE SURVEY AND VOTE

 

 

ETX Module Provides Long Life Cycle Solution

ADLINK Technology has continued its commitment and support to customers who have designs based on the ETX computer-on-module form factor. ETX is one of the earliest computer-on-module form factors. After more than two decades, its popularity is only second to COM Express when it comes to installed user base. With the recent discontinuation of the hugely popular Intel Atom processor N270, many customers are searching for an ETX module replacement to keep their systems up and running.

They are in need of an ETX drop-in solution at both hardware and software levels (Intel-to-Intel) with equivalent or improved performance and a better thermal envelope to simplify the transition. The problem is that customer’s current module suppliers may not have ETX on their roadmaps anymore. Since ETX is no longer a viable choice for completely new designs, many earlier manufacturers have moved on and dropped out of the ETX market.

ADLINK’s solution is the ETX-BT, based on the Intel Atom processor E3800 series SoC (formerly Bay Trail).  This Intel Atom product family is possibly the last processor that can fully support all ETX legacy interfaces: PATA IDE, ISA bus, PCI bus, serial/parallel ports, VGA and LVDS (Hsync/Vsync mode). The ETX-BT is available in both commercial (0°C to 60°C)  and Extreme Rugged (-40°C to +85°C) versions and has a life cycle of 10 years, keeping in line with Intel’s warranted life cycle for the Intel Atom processor E3800 series of 15 years from release.

ADLINK was a pioneer in the ETX form factor computer-on-module market, and continues to support its users in sustaining and extending the life of their existing ETX-based system.

ADLINK Technology | www.adlinktech.com

Linux-Driven Modules and SBC Tap i.MX8, i.MX8M and iMX8X

By Eric Brown

Phytec has posted product pages for three PhyCore modules, all of which support Linux and offer a -40°C to 85°C temperature range. The three modules, which employ three different flavors of i.MX8, include a phyCORE-i.MX 8X COM, which is the first product we’ve seen that uses the dual- or quad-core Cortex-A35 i.MX8X.

phyCORE-i.MX 8X (top) and phyCORE-i.MX 8M (bottom – not to scale) (click images to enlarge)

The phyCORE-i.MX 8 taps the high-end, hexa-core -A72 and -A53 i.MX8, including the i.MX8 QuadMax. The phyCORE-i.MX 8M, which uses the more widely deployed dual- or quad-core i.MX8M, is the only module that appears as part of an announced SBC: the sandwich-style phyBoard-Polaris SBC (shown). The phyCORE-i.MX 8 will also eventually appear on an unnamed, crowd-sourced Pico-ITX SBC.

phyCORE-i.MX 8 (left) and NXP i.MX8 block diagram (bottom)
(click images to enlarge)

Development-only carrier boards will be available for the phyCORE-i.MX 8X and phyCORE-i.MX 8. Evaluation kits based on the carrier boards and the phyBoard-Polaris will include BSPs with a Yocto Project based Linux distribution “with pre-installed and configured packages such as QT-Libs, OpenGL and Python.” Android is also available, and QNX, FreeRTOS and other OSes are available on request. BSP documentation will include a hardware manual, quickstart instructions, application guides, and software and application examples.

 

i.MX8M, i.MX8X, and i.MX8 compared (click image to enlarge)

The three modules are here presented in order of ascending processing power.

phyCore-i.MX 8X

The i.MX8X SoC found on the petite phyCORE-i.MX 8X module was announced with other i.MX8 processors in Oct. 2016 and was more fully revealed in Mar. 2017. The industrial IoT focused i.MX8X includes up to 4x cores that comply with Arm’s rarely used Cortex-A35 successor to the Cortex-A7 design.

phyCore-i.MX 8X (top) and block diagram (bottom)
(click images to enlarge)

The 28 nm fabricated, ARMv8 Cortex-A35 cores are claimed to draw about 33 percent less power per core and occupy 25 percent less silicon area than Cortex-A53. Phytec’s comparison chart shows the i.MX8X with 5,040 to 10,800 DMIPS performance, which is surprisingly similar to the 3,450 to 13,800 range provided by the Cortex-A53 based i.MX8M (see above).The i.MX8X SoC is further equipped with a single Cortex-M4 microcontroller, a Tensilica HiFi 4 DSP, and a multi-format VPU that supports up to 4K playback and HD encode. It uses the same Vivante GC7000Lite GPU found on the i.MX8M, with up to 28 GFLOPS.

i.MX8X block diagram
(click image to enlarge)

The i.MX8X features ECC memory support, reduced soft-error-rate (SER) technology, hardware virtualization, and other industrial and automotive safety related features. Crypto features listed for the phyCore-i.MX 8X COM include AES, 3DES, RSA, ECC Ciphers, SHA1/256, and TRNG.

PhyCore-i.MX7

Phytec’s 52 mm x 42 mm phyCore-i.MX 8X is only slightly larger than the i.MX7-based PhyCore-i.MX7, but the layout is different. The module supports all three i.MX8X models: the quad-core i.MX8 QuadXPlus and the dual-core i.MX8 DualXPlus and i.MX8 DualX, all of which can clock up to 1.2 GHz. The DualX model differs in that it has a 2-shader instead of 4-shader Vivante GPU.

The phyCore-i.MX 8X offers a smorgasbord of memories. In addition to the “128 kB multimedia,” and “64 kB Secure” found on the i.MX8X itself, the module can be ordered with 512 MB to 4 GB of LPDDR4 RAM and 64 MB to 256 MB of Micron Octal SPI/DualSPI flash. (Phytec notes that it is an official member of Micron’s Xccela consortium.) You can choose between 128 MB to 1 GB NAND flash or  4GB to 128 GB eMMC.

There’s no onboard wireless, but you get dual GbE controllers (1x onboard, 1x RGMII). You can choose between 2x LVDS and 2x MIPI-DSI. There are MIPI-CSI and parallel camera interfaces, as well as ESAI based audio.

Other I/O available through the 280 pins found on its two banks of dual 70-pin connectors include USB 3.0, USB OTG, PCI/PCIe, and up to 10x I2C. You also get 2x UART, 3x CAN, 6x A/D, and single PWM, keypad, or MMC/SD/SDIO (but only if you choose the eMMC over NAND). For SPI you get a choice of a single Octal connection or 2x “Quad SPI + 3 SPI” interfaces.

 

phyCore-i.MX 8X carrier board
(click image to enlarge)

The 3.3 V module supports an RTC, and offers watchdog and tamper features. Like all the new Phytec modules, you get -40°C to 85°C support. No details were available on the carrier shown in the image above.

phyCORE-i.MX 8M

The 55 mm x 40 mm phyCORE-i.MX 8M joins a growing number of Linux-driven i.MX8M modules including Compulab’s CL-SOM-iMX8, Emcraft’s i.MX 8M SOM, Innocom’s WB10, Seco’s SM-C12, SolidRun’s i.MX8 SOM, and the smallest of the lot to date: Variscite’s 55 x 30mm DART-MX8M. There are also plenty of SBCs to compete with the phyCORE-i.MX 8M-equipped phyBoard-Polaris SBC (see farther below), but like most of the COMs, most have yet to ship.

phyCORE-i.MX 8M top) and block diagram (bottom) (click images to enlarge)

The phyCORE-i.MX 8M supports the NXP i.MX8M Quad and QuadLite, both with 4x Cortex-A53 cores, as well as the dual-core Dual. All are clocked to 1.5 GHz. They all have 266MHz Cortex-M4F cores and Vivante GC7000Lite GPUs, but only the Quad and Dual models support 4Kp60, H.265, and VP9 video capabilities. (NXP also has a Solo model that we have yet to see, which offers a single -A53 core, a Cortex-M4F, and a GC7000nanoUltra GPU.)In addition to the i.MX8M SoC, which offers “128 KB + 32 KB” RAM and the same crypto features found on the i.MX8X, the module ships with the same memory features as the phyCore-i.MX 8X except that it lacks the SPI flash. Once again, you get 512 MB to  4 GB of LPDDR4 RAM and either 128 MB to 1 GB NAND flash or 4 GB to 128 GB eMMC. There is also SPI driven “Nand/QSPI” flash.

There’s a single GbE controller, and although not listed in the spec list, the product page says that precertified WiFi and Bluetooth BLE 4.2 are onboard and accompanied by antennas.

Multimedia support includes MIPI-DSI, HDMI 2.0, 2x MIPI-CSI, and up to 5x SAI audio. The block diagram also lists eDP, possibly as a replacement for HDMI.

Other interfaces expressed via the dual 200-pin connectors include 2x USB 3.0, 4x UART, 4x I2C, 4x PWM, and single SDIO and PCI/PCIe connections. SPI support includes 2x SPI and the aforementioned Nand/QSPI. The 3.3V module supports an RTC, watchdog, and tamper protections.

phyBoard-Polaris SBC

The phyCORE-i.MX 8M is also available soldered onto a carrier board that will be sold as a monolithic phyBoard-Polaris SBC. The 100 mm x 100 mm phyBoard-Polaris SBC features the Quad version of the phyCORE-i.MX 8M clocked to 1.3 GHz, loaded with 1 GB KPDDR4 and 8 GB eMMC. The SBC also adds a microSD slot.

phyBoard-Polaris SBC
(click image to enlarge)

The phyBoard-Polaris SBC is further equipped with single GbE, USB 3.0 and USB OTG ports. There’s also an RS-232 port and MIPI-DSI and SAID audio interfaces made available via A/V connectors. Dual MIPI-CSI interfaces are also onboard.A mini-PCIe slot and GPIO slot are available for expansion. The latter includes SPI, UART, JTAG, NAND, USB, SPDIF and DIO.

Other features include a reset button, RTC with coin cell, and JTAG via a debug adapter (PEB-EVAL). There’s a 12 V – 24 V input and adapter, and the board offers the same industrial temperature support as all the new Phytec modules.

phyCORE-i.MX 8

The phyCORE-i.MX 8, which is said to be “ideal for image and speech recognition,” is the third module we’ve seen to support NXP’s top-of-the-line, 64-bit i.MX8 series. The module supports all three flavors of i.MX8 while the other two COMs we’ve seen have been limited to the high-end QuadMax: Toradex’s Apalis iMX8 and iWave’s iW-RainboW-G27M.

phyCORE-i.MX 8 (top) and block diagram (bottom)
(click images to enlarge)

Like Rockchip’s RK3399, NXP’s hexa-core i.MX8 QuadMax features dual high-end Cortex-A72 cores clocked to up to 1.6 GHz plus four Cortex-A53 cores. The i.MX8 QuadPlus design is the same, but with only one Cortex-A72 core, and the quad has no -A72 cores.All three i.MX8 models provide two Cortex-M4F cores for real-time processing, a Tensilica HiFi 4 DSP, and two Vivante GC7000LiteXS/VX GPUs. The SoC’s “full-chip hardware-based virtualization, resource partitioning and split GPU and display architecture enable safe and isolated execution of multiple systems on one processor,” says Phytec.

The 73 mm x 45 mm phyCORE-i.MX 8 supports up to 8 GB LPDDR4 RAM, according to the product page highlights list, while the spec list itself says 1 GB to 64 GB. Like the phyCORE-i.MX 8X, the module provides 64 MB to 256 MB of Micron Octal SPI/DualSPI flash. There’s no NAND option, but you get 4 GB to 128 GB eMMC.

The phyCORE-i.MX 8 lacks WiFi, but you get dual GbE controllers. Other features expressed via the 480 connection pins include single USB 3.0, USB OTG, and PCIe 2.0 based SATA interfaces. Dual PCIe interfaces are also available

The module provides a 4K-ready HDMI output, 2x LVDS, and 2x MIPI-DSI for up 4x simultaneous HD screens. For image capture you get 2x MIPI-CSI and an HDMI input. Audio features are listed as “2x ESAI up to 4 SAI.”

The phyCORE-i.MX 8 is further equipped with I/O including 2x UART, 2x CAN, 2x MMC/SD/SDIO, 8x A/D, up to 19x I2C, and a PWM interface. For SPI, you get “up to 4x + 1x QSPI.” The module supports an RTC and offers industrial temperature support.

phyCORE-i.MX 8 carrier board (click image to enlarge)

In addition to the unnamed carrier board for the phyCORE-i.MX 8 module shown above, Phytec plans to produce a “Machine Vision and Camera kit” to exploit i.MX8 multimedia features including the VPU, the Vivante GPU’s Vulkan and OGL support, and interfaces including MIPI-DSI, MIPI-CSI, HDMI, and LVDS. In addition, the company will offer rapid prototyping services for customizing customer-specific hardware I/O platforms.Finally, Phytec is planning to develop a smaller, Pico-ITX form factor SBC based on the i.MX8 SoC, and it’s taking a novel approach to do so. The company has launched a Cre-8 community which intends to crowdsource the SBC. The company is seeking developers to join this alpha-stage project to contribute ideas. We saw no promises of open source hardware support, however.

Further information

[As of March 29] No availability information was provided for the phyCORE-i.MX 8X, phyCORE-i.MX 8M, or phyCORE-i.MX 8 modules, but the phyCORE-i.MX 8M-based phyBoard-Polaris is due in the third quarter. More information may be found in Phytec’s phyCORE-i.MX 8X, phyCORE-i.MX 8M, and phyCORE-i.MX 8 product pages as well as the phyBoard-Polaris SBC product page. More on development kits for all these boards may be found here.

This article originally appeared on LinuxGizmos.com on March 29.

Phytec issue a Press Release announcing these products on April 19.
UPDATE: “Early access program sampling for the phyCORE-i.MX8 and phyCORE-i.MX8M is planned for Q3 2018, with general availability expected in Q4 2018.”

Phytec | www.phytec.eu

Intel Coffee Lake H-Series Debut Processors Debut in Congatec and Seco Modules

By Eric Brown

Intel has rolled out new H-, M-, U- and T-series Intel Core and Xeon chips, expanding its line of 14 nm fabricated, 8th Gen Core Kaby Lake Refresh processors, code-named “Coffee Lake.” Of special interest are four new dual- and quad-core U-series chips with up to 2.7 GHz clock rates and 28 W TDPs, as well as four quad- and hexa-core H-series Core i5 and i7 processors and a pair of hexa-core M-series Xeon chips, all with 45 W TDP (total dissipated power).

Congatec Conga-TS370 (top) and Seco COMe-C08-BT6 (bottom) (click images to enlarge)

The hexa-core Core i7-8850H, quad-core Core i5-8400H, and hexa-core Xeon E-2176M are appearing in a pair of 125 mm x 95 mm COM Express Basic Type 6 announced by Congatec and Seco. Both the Conga-TS370 and Seco’s COMe-C08-BT6 are available with Linux or Windows 10, and support 0 to 60°C temperatures (see farther below).

Intel’s 8th Gen M- and H-series processors (click image to enlarge)

Intel launched its first round of 8th-Gen Kaby Lake Refresh “Coffee Lake” chips back in September. This fourth generation of its 14 nm fabricated Core chips — following Broadwell, Skylake, and Kaby Lake — offers relatively modest performance and power efficiency improvements.Like most of AMD’s new Ryzen Embedded V1000 SoCs, most of the Coffee Lake processors are double threaded, so four cores give you eight threads and six cores give you 12. The exception is a line of standard, desktop-oriented T-series chips with 35 W TDPs. The T-series models are all single-threaded except the top-of-the-line, hexa-core Core i7-8700T, clocked to 2.4 GHz /4.0 GHz.

Intel’s latest batch of U-series (top) and new T-series CPUs (bottom)
(click images to enlarge)

The latest batch of U-series processors give you more speed, but higher 28 W TDPs than the original batch. The initial U-series chips, which were used in recent Linux-based laptops from System76 and ZaReason, provide slightly faster quad- instead of dual-core designs with the same price and 15 W TDP as 7th-Gen “Kaby Lake” models. The first round of Coffee Lake chips also included some high-end models tuned to gaming, as well as the first hexa-core Core i5 and first quad-core Core i3 models.

Also today, Intel unveiled a new line of 300-series I/O chipsets that are based on the upcoming Cannon Lake PCH. The lineup includes a Q370 model that supports up to 6x USB 3.1 Gen2 ports, up to 24x PCIe 3.0 lanes, and Intel Wireless-AC for faster 802.11ac.

Intel’s original line of 8th Gen CPUs (top) and new 300-series I/O chipsets (bottom) (click images to enlarge)Intel added to the Coffee Lake parade with some gaming focused G-series chips that use a Radeon Vega GPU from rival AMD. The Core i7-8809G, which can be overclocked, as well as the fixed rate Core i7-8705G, are available in Intel NUC mini-PCs.Today’s media coverage emphasized Intel’s first mobile version of its gaming-oriented Core i9 design. The hexa-core Core i9-8950HK CPU uses thermal velocity boost” technology to jump from 2.9 GHz to 4.8 GHz.

The related H- and M-series processors used by Seco and Congatec include the Core i7-8850H, the fastest of the two hexa-core Core i7 models with 2. 6GHz /4.3 GHz performance. The i7-8850H offers a 9MB Intel Smart Cache and supports “partial” overclocking. The Core i5-8400H is the fastest of the two quad-core i5 models, with 2.5 GHz /4.2 GHz performance and an 8MB cache. The hexa-core, 2.7 GH z/4.4 GHz Xeon E-2176M with 12 MB cache is the slower of the two Xeon M-series chips. (The turbo speeds can only be achieved by one core at a time.)

All the models used by Congatec and Seco offer 45W TDPs and support Intel Optane memory and Intel VPro technology. As with other Coffee Lake processors, there are software patches to protect against Meltdown and Spectre vulnerabilities. However, a hardware fix will await the 10nm Cannon Lake generation.

The three models used by the Conga-TS370 and COMe-C08-BT6 modules are the:

  • Intel Core i7-8850H (6x 12-thread 14nm Coffee Lake cores at 2.6 GHz /4.3 GHz); 9 MB Cache, 45W TDP (35W cTDP)
  • Intel Core i5-8400H 4x 8-thread 14 nm Coffee Lake cores at 2.5 GHz /4.2 GHz); 8 MB Cache, 45W TDP (35W cTDP)
  • Intel Xeon E-2176M, 8850H (6x 12-thread 14 nm Coffee Lake cores at 2.7 GHz /4.4 GHz); 9 MB Cache, 45 W TDP (35W cTDP)

Intel claims that the six-core H-series and M-series modules offer between 45 to 50 percent more multi-thread and 15 to 25 percent more single-thread performance compared to 7th Gen “Kaby Lake” Core processors. The built-in Intel Gen9 LP graphics can manage up to 3x independent displays at once, with a resolution up to 4096 x 2304 at 60 Hz, 24 bpp. There’s support for DirectX 12 and OpenGL 4.5, as well as an H.265 / HEVC hardware transcoder.

Conga-TS370

Like Congatec’s 6th Gen Skylake based Conga-TS170 and 7th Gen Kaby Lake powered Conga-TS175, the Conga-TS370 uses the COM Express Type 6 Basic form factor. All common Linux operating systems, as well as the 64-bit versions of Microsoft Windows 10 and Windows 10 IoT are supported.

 

Conga-TS370 block diagram
(click image to enlarge)

The module offers up to 10-year availability, and targets applications including “high performance embedded and mobile systems, industrial and medical workstations, storage servers and cloud workstations, as well as media transcoding and edge computing cores,” says Congatec.Thanks to the Coffee Lake-H chips, the module supports Intel Optane memory, as well as Intel Software Guard extensions, Trusted Execution Engine, and Intel Platform Trust Technology. The Core processors use the new Intel PCH-H QM370 Series I/O chipset while the Xeon is paired with a CM246 Series controller.

You can load up to 32GB of  DDR4-2666 memory via dual sockets with optional ECC. There are 4x SATA III interfaces, as well as an Intel i219-LM GbE controller with AMT 12.0 support. Expansion features include a PEG x16 Gen3 interface and 8x PCIe Gen 3.0 lanes.

The integrated Intel UHD630 graphics supports up to three independent 4K displays via HDMI 1.4a, eDP 1.4, and DisplayPort 1.2. Dual-channel LVDS is also available as an alternative to eDP, and for the first time, you can switch between eDP to LVDS by software alone, says Congatec.

The highlighted feature enabled by Coffee Lake-H is its support for up to 4x USB 3.1 Gen 2 ports, which operate at up to 10 Gbps. The module also includes 8x USB 2.0 interfaces.

The Conga-TS370 is further equipped with LPC, I2C, SMBus, GPIO, SDIO, and dual UARTs. There’s also an HD Audio interface, TPM 2.0, and ACPI 4.0 with battery support. The Congatec Board Controller provides features including watchdog, non-volatile user storage, and backlight control.

Support services are available, along with a range of accessories and standardized or customized carrier boards and systems. A Conga-Teva2 carrier is in the works but is not yet documented.

COMe-C08-BT6

Seco’s COMe-C08-BT6 module, which follows it similarly Type 6, 6th Gen Skylake based COMe-B09-BT6, is designed for applications including gaming, signage, infotainment, HMI, biomedical devices, Industry 4.0, automation, and telco. There’s support for 64-bit Linux and Windows 10.

 

COMe-C08-BT6
(click image to enlarge)

Not surprisingly, the feature set is very similar to that of the Conga-TS370. You get up to 3 2GB of DDR4-2666 with ECC, 4x SATA 3.0 channels, and an Intel i219-LM GbE controller.The COMe-C08-BT6 has the same triple display and 4K support as the Congatec model. In this case you get DP, HDMI, and DVI DDI interfaces, as well as a choice of eDP, LVDS, or LVDS + VGA interfaces. HD Audio is also available.

Like the Conga-TS370, there are 4x USB 3.1 Gen 2 interfaces, 8x USB 2.0 links, a PEG x16 Gen3 interface, and 8x PCIe Gen 3.0 lanes. Other features include 2x UARTs, as well as SPI, I2C, SMBus, LPC, and GPIO. You also get a watchdog, optional TPM 2.0, thermal and fan management signals, and 12 V or optional 5 V DC input.

CCOMe-965 carrier (top) and block diagram (bottom)
(click images to enlarge)The COMe-C08-BT6 is available with Seco’s CCOMe-965 Mini-ITX carrier board, which also supports other Seco Type 6 modules such as the COMe-B09-BT6 and Ryzen V1000 based COMe-B75-CT6. There’s also a Cross Platform Development Kit that includes the CCOMe-965, along with HDMI and DisplayPort cables, and is said to support ARM-based Type 6 COMs in addition to x86.

CCOMe-C30 carrier (top) and block diagram (bottom)
(click images to enlarge)One final development option is an upcoming, 3.5-inch form factor CCOMe-C30 board that features a DP++ port, 2x mini-DP++ ports, and LVDS and eDP connections. The 146 mm x 102 mm board has dual M.2 sockets, dual GbE ports, and SATA and microSD slots. You also get 2x USB 3.0 and 2x USB 2.0 ports, plus 4x serial headers, among other features.Further information

No pricing or availability information was provided for the Congatec Conga-TS370 or Seco COMe-C08-BT6 Type 6 modules. More on Congatec’s Conga-TS370 module may be found in the Conga-TS370 announcement and product pages.

More on Seco’s COMe-C08-BT6 may be found on the COMe-C08-BT6 product page.

Intel’s latest Intel Coffee Lake processors should start shipping in volume by the end of the month. More information may be found on Intel’s 8th Gen Intel Core announcement page.

This article originally appeared on LinuxGizmos.com on April 3.

Congatec | www.congtatec.com

Seco | www.seco.com

Commell Launches its First ARM-Based Pico-ITX

By Eric Brown

Commell has announced the LP-150, a Rockchip RK3128 based Pico-ITX SBC that appears to be its first ARM-based embedded board of any kind. The 100 mm x 72 mm LP-150 is the only ARM-based SBC out of the many dozens of mostly Intel-based boards listed on Commell’s SBC page.

Shipping with Android 4.4.4, but also supporting Linux, the LP-150 is intended primarily for imaging, machine vision and digital signage applications. Other Commell Pico-ITX SBCs include its Intel Braswell based LP-176.

Commell LP-150

Rockchip’s quad-core, Cortex-A7 RK3128 hasn’t seen as much uptake in the embedded world as the quad -A17 RK3288, which is found on hacker boards such as the Firefly-RK3288 Reload or the high-end, hexa-core RK3399, which has appeared on numerous recent products such as OpenEmbed’s em3399 module or Aaeon’s RICO-3399 PICO-ITX SBC. The only RK3128-based SBC we can recall is the open spec Firefly-FirePrime S.

The LP-150 SBC has a fairly modest feature set, with only 512 MB DDR3. Yet, it offers a few features you don’t typically find on x86 Pico-ITX SBCs like eMMC storage (8GB) and built-in Wi-Fi. Media features include an HDMI 1.4 port limited to HD resolution and an LVDS interface with capacitive touchscreen support. You also get CVBS inputs and outputs and audio I/O headers.

The LP-150 is further equipped with a GbE port and USB 2.0 host and OTG ports. There is also an RTC with battery, and a smattering of RS-232, UART, and GPIO interfaces.

Specifications listed for the Commell LP-150 include:

  • Processor — Rockchip RK3128 (4x Cortex-A7 @ 1.3 GHz); Mali-400 MP2 GPU with OpenGL ES1.1 and 2.0, OpenVG1.1
  • Memory — 512 MB DDR3
  • Storage — 8 GB eMMC; microSD slot
  • Display:
    • HDMI 1.4 port for up to 1080p
    • Single-channel 18/24-bit LVDS for up to 1280 x 720 displays or up to 1024 x 600 cap. touchscreens
    • LCD/LVDS panel and inverter connectors
    • CVBS in/out
  • Wireless — Wi-Fi with SMA antenna
  • Networking — Gigabit Ethernet port (RTL8211E)
  • Other I/O:
    • USB 2.0 host port with support for 4-port hub
    • USB 2.0 OTG port
    • 2x RS232 interfaces
    • 3x UART
    • Audio line-out, mic-in headers (Rockchip codec)
    • GPIO header
  • Other features — Power, recovery, reset buttons; RTC with lithium battery; LED
  • Operating temperature — 0 to 70°C
  • Power — DC input 5 V
  • Dimensions — 100 mm x 72 mm (Pico-ITX)
  • Operating system — Android 4.4.4; Linux also supported

Further information

No pricing or availability information was provided for the LP-150. More information may be found on Commell’s LP-150 product page.

Commell |  www.commell.com.tw

This article originally appeared on LinuxGizmos.com on March 6.

Movidius AI Acceleration Technology Comes to a Mini-PCIe Card

By Eric Brown

UP AI Core (front)

As promised by Intel when it announced an Intel AI: In Production program for its USB stick form factor Movidius Neural Compute Stick, Aaeon has launched a mini-PCIe version of the device called the UP AI Core. It similarly integrates Intel’s AI-infused Myriad 2 Vision Processing Unit (VPU). The mini-PCIe connection should provide faster response times for neural networking and machine vision compared to connecting to a cloud-based service.

UP AI Core (back)

The module, which is available for pre-order at $69 for delivery in April, is designed to “enhance industrial IoT edge devices with hardware accelerated deep learning and enhanced machine vision functionality,” says Aaeon. It can also enable “object recognition in products such as drones, high-end virtual reality headsets, robotics, smart home devices, smart cameras and video surveillance solutions.”

 

 

UP Squared

The UP AI Core is optimized for Aaeon’s Ubuntu-supported UP Squared hacker board, which runs on Intel’s Apollo Lake SoCs. However, it should work with any 64-bit x86 computer or SBC equipped with a mini-PCIe slot that runs Ubuntu 16.04. Host systems also require 1GB RAM and 4GB free storage. That presents plenty of options for PCs and embedded computers, although the UP Squared is currently the only x86-based community backed SBC equipped with a Mini-PCIe slot.

Myriad 2 architecture

Aaeon had few technical details about the module, except to say it ships with 512MB of DDR RAM, and offers ultra-low power consumption. The UP AI Core’s mini-PCIe interface likely provides a faster response time than the USB link used by Intel’s $79 Movidius Neural Compute Stick. Aaeon makes no claims to that effect, however, perhaps to avoid

Intel’s Movidius
Neural Compute Stick

disparaging Intel’s Neural Compute Stick or other USB-based products that might emerge from the Intel AI: In Production program.

It’s also possible the performance difference between the two products is negligible, especially compared with the difference between either local processing solutions vs. an Internet connection. Cloud-based connections for accessing neural networking services suffer from reduced latency, network bandwidth, reliability, and security, says Aaeon. The company recommends using the Linux-based SDK to “create and train your neural network in the cloud and then run it locally on AI Core.”

Performance issues aside, because a mini-PCIe module is usually embedded within computers, it provides more security than a USB stck. On the other hand, that same trait hinders ease of mobility. Unlike the UP AI Core, the Neural Compute Stick can run on an ARM-based Raspberry Pi, but only with the help of the Stretch desktop or an Ubuntu 16.04 VirtualBox instance.

In 2016, before it was acquired by Intel, Movidius launched its first local-processing version of the Myriad 2 VPU technology, called the Fathom. This Ubuntu-driven USB stick, which miniaturized the technology in the earlier Myriad 2 reference board, is essentially the same technology that re-emerged as Intel’s Movidius Neural Compute Stick.

UP AI Core, front and back

Neural network processors can significantly outperform traditional computing approaches in tasks like language comprehension, image recognition, and pattern detection. The vast majority of such processors — which are often repurposed GPUs — are designed to run on cloud servers.

AIY Vision Kit

The Myriad 2 technology can translate deep learning frameworks like Caffe and TensorFlow into its own format for rapid prototyping. This is one reason why Google adopted the Myriad 2 technology for its recent AIY Vision Kit for the Raspberry Pi Zero W. The kit’s VisionBonnet pHAT board uses the same Movidius MA2450 chip that powers the UP AI Core. On the VisionBonnet, the processor runs Google’s open source TensorFlow machine intelligence library for neural networking, enabling visual perception processing at up to 30 frames per second.

Intel and Google aren’t alone in their desire to bring AI acceleration to the edge. Huawei released a Kirin 970 SoC for its Mate 10 Pro phone that provides a neural processing coprocessor, and Qualcomm followed up with a Snapdragon 845 SoC with its own neural accelerator. The Snapdragon 845 will soon appear on the Samsung Galaxy S9, among other phones, and will also be heading for some high-end embedded devices.

Last month, Arm unveiled two new Project Trillium AI chip designs intended for use as mobile and embedded coprocessors. Available now is Arm’s second-gen Object Detection (OD) Processor for optimizing visual processing and people/object detection. Due this summer is a Machine Learning (ML) Processor, which will accelerate AI applications including machine translation and face recognition.

Further information

The UP AI Core is available for pre-order at $69 for delivery in late April. More information may be found at Aaeon’s UP AI Core announcement and its UP Community UP AI Edge page for the UP AI Core.

Aaeon | www.aaeon.com

This article originally appeared on LinuxGizmos.com on March 6.