SBC Showcases Qualcomm’s 10 nm, Octa-core QCS605 IoT SoC

By Eric Brown

In April, Qualcomm announced its QCS605 SoC, calling it “the first 10nm FinFET fabricated SoC purpose built for the Internet of Things.” The octa-core Arm SoC is available in an Intrinsyc Open-Q 605 SBC with full development kit with a 12V power supply is open for pre-orders at $429. The products will ship in early December.

 
Open-Q 605, front and back
(click images to enlarge)
The fact that Qualcomm is billing the high-end QCS605 as an IoT SoC reveals how demand for vision and AI processing on the edge is broadening the IoT definition to encompass a much higher range of embedded technology. The IoT focus is also reinforced by the lack of the usual Snapdragon branding. The QCS605 is accompanied by the Qualcomm Vision Intelligence Platform, a set of mostly software components that includes the Qualcomm Neural Processing SDK and camera processing software, as well as the company’s 802.11ac WiFi and Bluetooth connectivity and security technologies.

The QCS605 can run Linux or Android, but Intrinsyc supports its Open-Q 605 board only with Android 8.1.

Intrinsyc also recently launched an Open-Q 624A Development Kit based on a new Open-Q 624A SOM (see farther below).

Qualcomm QCS605 and Vision Intelligence Platform

The QCS605 SoC features 8x Kryo 300 CPU cores, two of which are 2.5GHz “gold” cores that are equivalent to Cortex-A75. The other six are 1.7GHz “silver” cores like the Cortex-A55 — Arm’s more powerful follow-on to Cortex-A53.

The QCS605 also integrates an Adreno 615 GPU, a Hexagon 685 DSP with Hexagon vector extensions (“HVX”), and a Spectra 270 ISP that supports dual 16-megapixel image sensors. Qualcomm also sells a QCS603 model that is identical except that it offers only 2x of the 1.7GHz “Silver” cores instead of six.

Qualcomm sells the QCS605 as part of a Vision Intelligence Platform — a combination of software and hardware starting with a Qualcomm AI Engine built around the Qualcomm Snapdragon Neural Processing Engine (NPE) software framework. The NPE provides analysis, optimization, and debugging tools for developing with Tensorflow, Caffe, and Caffe2 frameworks. The AI Engine also includes the Open Neural Network Exchange interchange format, the Android Neural Networks API, and the Qualcomm Hexagon Neural Network library, which together enable the porting of trained networks.

The Vision Intelligence Platform running on the QCS605 delivers up to 2.1 TOPS (trillion operations per second) of compute performance for deep neural network inferences, claims Qualcomm. The platform also supports up to 4K60 resolution or 5.7K at 30fps and supports multiple concurrent video streams at lower resolutions.

Other features include “staggered” HDR to prevent ghost effects in high-dynamic range video. You also get advanced electronic image stabilization, de-warp, de-noise, chromatic aberration correction, and motion compensated temporal filters in hardware.

Inside the Open-Q 605 SBC

Along with the Snapdragon 600 based Open-Q 600, the Open-Q 605 is the only Open-Q development board that Intrinsyc refers to as an SBC. Most Open-Q kits are compute modules or sandwich-style carrier board starter kits based on Intrinsyc modules equipped with Snapdragon SoCs, such as the recent, Snapdragon 670 based Open-Q 670 HDK.


Open-Q 605 
(click image to enlarge)
The 68 x 50mm Open-Q 605 ships with an eMCP package with 4GB LPDDR4x RAM and 32GB eMMC flash, and additional storage is available via a microSD slot. Networking depends on the 802.11ac (WiFi 5) and Bluetooth 5.x radios. There’s also a Qualcomm GNSS receiver for location and 3x U.FL connectors.

The only real-world coastline port is a USB Type-C that supports DisplayPort 1.4 with 4K@30fps support. If you’d rather use the Type-C port for USB or charging a user-supplied Li-Ion battery, you can turn to an HD-ready MIPI DSI interface with touch support. You also get 2x MIPI-CSI for dual cameras, as well as 2x analog audio.

The Open-Q 605 has a 76-pin expansion header for other interfaces, including an I2S/SLIMBus digital audio interface. The board runs on a 5-15V DC input and offers an extended -25 to 60°C operating range.

Specifications listed for the Open-Q 605 SBC include:

  • Processor — Qualcomm QCS605 with Vision Intelligence Platform (2x up to 2.5GHz and 6x up to 1.7GHz Krait 300 cores); Adreno 615 GPU; Hexagon 685 DSP; Spectra 270 ISP; Qualcomm AI Engine and other VIP components
  • Memory/storage — 4GB LPDDR4X and 32GB eMMC flash in combo eMCP package; microSD slot.
  • Wireless:
    • 802.11b/g/n/ac 2×2 dual-band WiFi (Qualcomm WCN3990) with planned FCC/IC/CE certification
    • Bluetooth 5.x
    • Qualcomm GNSS (SDR660G) receiver with Qualcomm Location Suite Gen9 VT
    • U.FL antenna connectors for WiFi, BT, GNSS
  • Media I/O:
    • DisplayPort 1.4 via USB Type-C up to 4K@30 with USB data concurrency (USB and power)
    • MIPI DSI (4-lane) with I2C touch interface on flex cable connector for up to 1080p30
    • 2x MIPI-CSI (4-lane) with micro-camera module connectors
    • 2x analog mic I/Ps, speaker O/P, headset I/O
    • I2S/SLIMBus digital audio interface with 2x DMIC ports (via 76-pin expansion header)
  • Expansion — 76-pin header (multiple SPI, I2C, UART, GPIO, and sensor I/O; digital and analog audio I/O, LED flash O/P, haptic O/P, power output rails
  • Other features — 3x LEDs; 4x mounting holes; optional dev kit with quick start guide, docs, SW updates
  • Operating temperature — -25 to 60°C
  • Power — 5-15V DC jack and support for user-supplied Li-Ion battery with USB Type-C charging; PM670 + PM670L PMIC; 12V supply with dev kit
  • Dimensions — 68 x 50 x 13mm
  • Operating system — Android 8.1 Oreo

Open-Q 624A
Development Kit

Open-Q 624A Development Kit

Back in May, Google preannounced the Open-Q 624A Development Kit as an official Android Things 1.0 development board along with Intrinsyc’s Snapdragon 212 based Open-Q 212A, Innocomm’s i.MX8M based WB10-AT, and a MediaTek MT8516 development platform. Now, Intrinsyc is pitching the Open-Q 624A Development Kit, as well as the Open-Q 624A SOM module it’s based on, as an Android 8.0 platform aimed at the home hub market. There is no longer any mention of Android Things.

The Open-Q 624A SOM offers 2GB RAM, 4GB eMMC, WiFi-ac, BT 4.2, and an octa-core -A53 Qualcomm Snapdragon 624 SoC based on the Snapdragon 625. The kit is equipped with a USB 3.0 Type-C port, 2x USB host ports, micro-USB client and debug ports, MIPI-CSI and MIPI-DSI interfaces, sensor expansion and haptic output, and an optional GPS receiver. You also get extensive audio features, including I2S/SLIMBUS headers.

Available for $595, the sandwich style kit will ship in mid-December. For more details, see our earlier Android Things development board report.

Further information

The Open-Q 605 SBC is available for pre-order in the full Development Kit version, which costs $429 and ships in early December. The SBC will also be sold on its own at an undisclosed price. More information may be found in Intrinsyc’s Open-Q 605 announcement, as well as the product page and shopping page.

This article originally appeared on LinuxGizmos.com on November 14.

Intrinsyc | www.intrinsyc.com

Bluetooth Smart SoCs Links Wearables to Apps for WeChat

Dialog Semiconductor recently announced its support for WeChat’s communications protocol with the launch of its WeChat SDK. With the kit, you can quickly add Bluetooth connectivity between WeChat apps and wearables and other IoT devices. Dialog DA14580 Dialog’s development kit is available now and includes a protocol stack for the WeChat communication layer. The SDK—which is based on the DA1458x family of SmartBond SoCs—enables you to reduce the overall development time for connecting their products wirelessly to WeChat apps. Your users can control wearable devices via the app and share information via the platform.

DA1458x SoCs combine a Bluetooth low-energy radio with an ARM Cortex-M0 application processor. With intelligent power management circuitry and accessible processor resources via 32 GPIOs,you can build fully hosted applications.

The SmartBond WeChat SDK enables efficient coding and comes with SmartSnippets software development environment, which is based on Keil µVision tools.

Source: Dialog Semiconductor

New JukeBlox Wi-Fi Platform for Streaming Audio

Microchip Technology’s fourth-generation JukeBlox platform enables product developers to build low-latency systems, such as wireless speakers, sound bars, AV receivers, micro systems, and more. The JukeBlox 4 Software Development Kit (SDK) in combination with the CY920 Wi-Fi & Bluetooth Network Media Module features dual-band Wi-Fi technology, multi-room features, AirPlay and DLNA connectivity, and integrated music services.

Microchip-JukeBlox-Wifi

Streaming audio with JukeBlox

The CY920 module is based on Microchip’s DM920 Wi-Fi Network Media Processor, which features 2.4- and 5-GHz 802.11a/b/g/n Wi-Fi, high-speed USB 2.0 and Ethernet connectivity. By using the 5-GHz band, speakers aren’t impacted by the RF congestion found in the 2.4-GHz band.

The DM920 processor also features integrated dual 300-MHz DSP cores that can reduce or eliminate the need for costly standalone DSP chips. An PC-based GUI simplifies the use of a predeveloped suite of standard speaker-tuning DSP algorithms, including a 15-band equalizer, multiband dynamic range compression, equalizer presets, and a variety of filter types. Even if you don’t have DSP coding experience, you can implement DSP into your designs.

JukeBlox 4 enables you to directly stream cloud-based music services, such as Spotify Connect and Rhapsody, while using mobile devices as remote controls. Mobile devices can be used anywhere in the Wi-Fi network without interrupting music playback. In addition, JukeBlox technology offers cross-platform support for iOS, Android, Windows 8, and Mac, along with a complete range of audio codecs and ease-of-use features to simplify network setup.

The JukeBlox 4 SDK, along with the JukeBlox CY920 module, is now available for sampling and volume production.

Source: Microchip Technology

SDK for OpenCL Dev Flow

Altera Corp. has simplified a programmer’s ability to accelerate algorithms in FPGAs. The Altera SDK for OpenCL version 14.0 includes a programmer-familiar rapid prototyping design flow that enables users to prototype designs in minutes on an FPGA accelerator board. Altera, along with its board partners, further accelerate the development of FPGA-based applications by offering reference designs, reference platforms and FPGA development boards that are supported by Altera’s OpenCL solution. These reference platforms also streamline the development of custom FPGA accelerators to meet specific application requirements.

Altera is the only company to offer a publicly available, OpenCL conformant software development kit (SDK). The solution allows programmers to develop algorithms with the C-based OpenCL language and harness the performance and power efficiencies of FPGAs. A rapid prototyping design flow included in the Altera SDK for OpenCL version 14.0 allows OpenCL kernel code to be emulated, debugged, optimized, profiled and re-compiled to a hardware implementation in minutes. The re-compiled kernels can be tested and run on an FPGA immediately, saving programmers weeks of development time.

Altera and its board partners further simplify the experience of getting applications up and running using FPGA accelerators by offering a broad selection of Altera-developed reference platforms, reference designs and FPGA accelerator boards. Altera provides a variety of design examples that demonstrate how to describe applications in OpenCL, including OPRA FAST Parser for finance applications, JPEG decoder for big data applications and video downscaling for video applications.

Design teams that want to create custom solutions that feature a unique set of peripherals can create their own custom FPGA accelerators and save significant development time by using Altera-developed reference platforms. The reference platforms include an SoC platform for embedded applications, a high-performance computing (HPC) platform and a low-latency network enabled platform which utilizes IO Channels.

One notable enhancement is production support for I/O Channels that allow streaming data into and out of the FPGA as well as kernel channels allowing the result reuse from one kernel to another in a hardware pipeline for significantly higher performance and throughput with little to no host and memory interaction. Another enhancement is production support for single-chip SoC solutions (Cyclone V SoC and Arria V SoC), where the host is an embedded ARM core processor integrated in the FPGA accelerator.

Altera’s SDK for OpenCL allows programmers to take OpenCL code and rapidly exploit the massively parallel architecture of an FPGA. Programmers targeting FPGAs achieve higher performance at significantly lower power compared to alternative hardware architectures, such as GPUs and CPUs. On average, FPGAs deliver higher performance at one-fifth the power of a GPU. Altera’s OpenCL solutions are supported by third-party boards through the Altera Preferred Board Partner Program for OpenCL. Visit www.altera.com/opencl.

The Altera SDK for OpenCL is currently available for download on Altera’s website (www.altera.com/products/software/opencl/opencl-index.html). The annual software subscription for the SDK for OpenCL is $995 for a node-locked PC license. For additional information about the Altera Preferred Board Partner Program for OpenCL and its partner members, or to see a list of all supported boards and links to purchase, visit the OpenCL section on Altera’s website.

[Source: Altera Corp.]

Arduino USB Host Shield

The Arduino USB Host Shield allows you to connect a USB device to your Arduino board. The Arduino USB Host Shield is based on the MAX3421E, which is a USB peripheral/host controller containing the digital logic and analog circuitry necessary to implement a full-speed USB peripheral or a full-/low-speed host compliant to USB specification rev 2.0.ArduinoHostshield

The shield is TinkerKit compatible, which means you can quickly create projects by plugging TinkerKit modules onto the board. The following device classes are supported by the shield:

  • HID devices: keyboards, mice, joysticks, etc.
  • Game controllers: Sony PS3, Nintendo Wii, Xbox360
  • USB to serial converters: FTDI, PL-2303, ACM, as well as certain cell phones and GPS receivers
  • ADK-capable Android phones and tables
  • Digital cameras: Canon EOS, Powershot, Nikon DSLRs and P&S, as well as generic PTP
  • Mass storage devices: USB sticks, memory card readers, external hard drives, etc.
  • Bluetooth dongles

For information on using the shield with the Android OS, refer to Google’s ADK documentation. Arduino communicates with the MAX3421E using the SPI bus (through the ICSP header). This is on digital pins 10, 11, 12, and 13 on the Uno and pins 10, 50, 51, and 52 on the Mega. On both boards, pin 10 is used to select the MAX3421E.

[Source: Arduino website via Elektor]