Qualcomm Robotics RB5/RB6 - SoM Overview

Qualcomm Robotics RB5/RB6
Introduction
SoM Overview Carrier Boards Qualcomm Robotics RB5 developer kit Qualcomm Robotics RB6 platform
Software Development Support
Boot Devices Thundercomm SDK Downloading SDK Installing SDK Generating Image and flashing Linaro Getting Started Flashing Bootloader Flashing Boot Image and Rootfs Special / Debug Modes Setting up board in EDL Mode Setting up board in Fastboot Mode
Development in the Board
Getting into the Board Using ADB Using Serial Console Using SSH Device Setup WiFi setup HDMI setup Update Ubuntu Packages
Build from source
Base Image Customize Image
Capture Subsystem
Hardware Capture Components Software Capture Components Building your own camera sensor driver
Image Processing and Acceleration Software
GPU Profiling Fast CV Installing Hexagon SDK Installing Fast CV Building your own Fast CV application
GStreamer Pipelines
Capture and Display Video Encoding Video Decoding
AI Acceleration
Neural Processing SDK Downloading Requirements Setup SDK Environment Install qtimlesnpe Example Project TensorFlow Example pipeline
Contact Us

This section provides an overview of the Robotics RB5/RB6 SoM components and their capabilities.

Error creating thumbnail: Unable to save thumbnail to destination

Figure 1: Qualcomm Robotics RB5 SOM

The Qualcomm Robotics RB5/RB6 System-on-Module (SoM) is a PCB containing the platform's memory, power management IC, communication interfaces and the powerful heterogeneous computing architecture QRB5165 SoC encapsulating the CPU, GPU, DSP AI engine, ISP, Neural Processing Unit (NPU) and Secure Processing Unit for security. Figure 1 shows a picture of the Robotics RB5 SoM.

Table 1 shows a summary of the hardware characteristics of the SoM:

Table 1: Qualcomm Robotics RB5/RB6 SoM - Hardware Summary

Feature	Qualcomm Robotics RB5/RB6
GPU	Adreno GPU 650
CPU	8-core Qualcomm® Kryo™ 585 built on Arm® Cortex® technology, 64-bit CPU
CPU Max Freq	2.842 GHz
Memory	Four-channel PoP high-speed memory – 8 GB LPDDR5 or LPDDR4X SDRAM (4 × 16-bit) designed for a 2750 MHz (LPDDR5) or 2133 MHz (LPDDR4X) clock and system cache. Memory density up to 16GB
Storage	UFS 3.1 gear 4 2 lanes, 128GB
Wireless Connectivity	2x WLAN 2 802.11ax with DBS and Bluetooth® 5.1
Digital Signal Processing	Compute Hexagon DSP v66 at 1.5 GHz with 4x Hexagon Vector eXtensions with an Hexagon Tensor Accelerator delivering 15 TOPS of AI performance (70-100 TOPS in RB6).
Neural Processing Unit	NPU230 for high-performance machine learning
Image Signal Processing	Qualcomm Spectra 480 ISP to support up to 12 cameras by D-PHY and 18 cameras by C-PHY (7 concurrent). It offers six CSI 4-lane interfaces with either D-PHY or C-PHY: D-PHY v1.2: 2.5 Gbps/lane C-PHY v1.2: 10.26 Gbps/T
Secure Processing Unit	SPU 240
Encoder	Adreno 665 Video Processing Unit: Up to 4K120/8K30 Native support for: H.265 Main 10, H.265 Main, H.264 High, and VP8
Decoder	Adreno 665 Video Processing Unit: Up to 4K240/8K60 Native support for: H.265 Main 10, H.265 Main, H.264 High, VP9 profile 2, VP8, and MPEG-2

Graphics Processing Unit (GPU)

The Qualcomm® Adreno™ 650 GPU has the following characteristics:

• Supports the following APIs:
  • For graphics: OpenGL ES 3.2, Vulkan and VulkanCompute, DX12.x
  • For computing: OpenCL 2.0 full profile, DirectCompute

Central Processing Unit (CPU)

The Qualcomm® Kryo™ 585 CPU's cores have the following characteristics:

• 4 high-performance Kryo Gold cores
• 4 low-power Kryo Silver cores

Digital Signal Processor (DSP)

The Hexagon DSP is more efficient in power consumption than the CPU. It is conformed by several DSPs such as the compute DSP (cDSP) which is preferable for compute-intensive tasks such as, video enhancements, virtual reality applications and camera streaming; it also contains an audio DSP (aDSP) and sensor DSP. The Hegaxon DSP comes with an Hexagon Tensor Accelerator (15 TOPS) for fixed-point deep convolutional neural networks and Hexagon Scalar Accelerator to support AI processing and computer vision (CV). The cDSP contains the Hexagon vector eXtensions (HVX) which performs fixed-point vector operations.

Neural Processing Unit (NPU)

The NPU 230 for neural networks processing in high-performance machine learning. Used along with the Hegaxon DSP.

Image Signal Processor (ISP)

The Qualcomm® Spectra 480 ISP can process 2 Gigapixels per second with the quality of 200 megapixel photos, record video at 8K and 4K HDR video capture. Has the following characteristics:

• Real-time sensor input resolution: Up to two 25MP and five 2MP cameras
• 64 MP 30 fps ZSL with a dual ISP

Secure processing Unit (SPU)

The SPU 240 comes with several features such as the following:

• Camera Security
• Crypto Engine
• Cryptographic Accelerator
• Qualcomm Trusted Execution Environment
• Secure Boot
• Qualcomm® Crypto Engine Core is FIPS 140-2 certified.