3. Foundational Components

• 3.1. U-Boot
  • 3.1.1. User’s Guide
    • 3.1.1.1. General Information
      • 3.1.1.1.1. Getting the U-Boot Source Code
      • 3.1.1.1.2. Build U-Boot
      • 3.1.1.1.3. Image Formats
      • 3.1.1.1.4. Boot Flow
      • 3.1.1.1.5. U-Boot Environment
      • 3.1.1.1.6. Available RAM for image download
      • 3.1.1.1.7. Device Trees
      • 3.1.1.1.8. SRAM memory Layout during R5 SPL bootloader stage
    • 3.1.1.2. USB Device Firmware Upgrade (DFU)
      • 3.1.1.2.1. USB Peripheral boot mode (SPL-DFU support)
    • 3.1.1.3. Network (Wired or USB Client)
      • 3.1.1.3.1. Booting U-Boot from the network
      • 3.1.1.3.2. Multiple Interfaces
      • 3.1.1.3.3. Network configuration via DHCP
      • 3.1.1.3.4. Manual network configuration
      • 3.1.1.3.5. Disabling Gigabit Phy Advertising
      • 3.1.1.3.6. Booting Linux from the network
      • 3.1.1.3.7. Sample Script for AM65 SR1
      • 3.1.1.3.8. Sample Script for AM65 SR2
    • 3.1.1.4. NAND
      • 3.1.1.4.1. Erasing, Reading and Writing to/from NAND partitions
      • 3.1.1.4.2. Writing to NAND via DFU
      • 3.1.1.4.3. NAND Boot
      • 3.1.1.4.4. Booting Kernel and Filesystem from NAND
      • 3.1.1.4.5. U-Boot Environment in NAND
    • 3.1.1.5. SD, eMMC and USB
      • 3.1.1.5.1. Partitioning eMMC from U-Boot
      • 3.1.1.5.2. Updating an SD card from a host PC
      • 3.1.1.5.3. Updating an SD card or eMMC using DFU
      • 3.1.1.5.4. Booting Linux from SD card or eMMC
      • 3.1.1.5.5. Booting tiboot3.bin, tispl.bin and u-boot.img from eMMC boot partition (For K3 class of SoCs)
      • 3.1.1.5.6. Booting to U-Boot prompt from USB storage
      • 3.1.1.5.7. Booting Linux from USB storage
      • 3.1.1.5.8. Steps for working around SD card issues
    • 3.1.1.6. SPI
      • 3.1.1.6.1. Writing to SPI from U-Boot
      • 3.1.1.6.2. Booting from SPI
    • 3.1.1.7. OSPI/QSPI NOR/NAND
    • 3.1.1.8. NOR
      • 3.1.1.8.1. Writing to NOR from U-Boot
      • 3.1.1.8.2. Booting from NOR
    • 3.1.1.9. UART
      • 3.1.1.9.1. Booting U-Boot from the console UART
    • 3.1.1.10. SATA
      • 3.1.1.10.1. Viewing SATA Devices
      • 3.1.1.10.2. Viewing Partitions
      • 3.1.1.10.3. Identifying Partition Filesystem Type
      • 3.1.1.10.4. Viewing, Reading and Writing to Partition
      • 3.1.1.10.5. View Partition Contents
    • 3.1.1.11. UFS
    • 3.1.1.12. DDR3 ECC
      • 3.1.1.12.1. DDR3 ECC in Keystone-II
        • 3.1.1.12.1.1. DDR3 ECC Handling
  • 3.1.2. Troubleshooting
    • 3.1.2.1. U-Boot Debug Tips
    • 3.1.2.2. Debugging SPL in CCS
      • 3.1.2.2.1. Step 1: Downloading CCS
      • 3.1.2.2.2. Step 2: Creating a Target Configuration File
      • 3.1.2.2.3. Step 3: Loading Symbol files
    • 3.1.2.3. Debugging SPL in OpenOCD
    • 3.1.2.4. Loading Uboot through CCS
      • 3.1.2.4.1. Step 1: Software Downloads
      • 3.1.2.4.2. Step 2: Testing Hardware Connection
      • 3.1.2.4.3. Step 3: Building the SPL/Uboot images
      • 3.1.2.4.4. Step 4: CCS Configuration
      • 3.1.2.4.5. Step 5: Change ARM mode
      • 3.1.2.4.6. Step 6: Loading SPL
      • 3.1.2.4.7. Step 7: Loading Uboot Image
    • 3.1.2.5. U-Boot Splash Screen
• 3.2. Kernel
  • 3.2.1. Users Guide
    • 3.2.1.1. Overview
    • 3.2.1.2. Getting the Kernel Source Code
    • 3.2.1.3. Preparing to Build
      • 3.2.1.3.1. Compiler
      • 3.2.1.3.2. Cleaning the Kernel Sources
    • 3.2.1.4. Configuring the Kernel
      • 3.2.1.4.1. Using Default Configurations
      • 3.2.1.4.2. Customizing the Configuration
    • 3.2.1.5. Compiling the Sources
      • 3.2.1.5.1. Compiling the Kernel
      • 3.2.1.5.2. Compiling the Device Tree Binaries
      • 3.2.1.5.3. Compiling the Kernel Modules
    • 3.2.1.6. Creating the kernel fitImage for high security device / GP devices
      • 3.2.1.6.1. Describing FIT source
      • 3.2.1.6.2. Generating the fitImage
      • 3.2.1.6.3. Build uboot again
    • 3.2.1.7. Installing the Kernel
      • 3.2.1.7.1. Installing the Kernel Image and Device Tree Binaries
      • 3.2.1.7.2. Installing the Kernel Modules
  • 3.2.2. Kernel Release Notes
    • 3.2.2.1. Build Information
    • 3.2.2.2. Generic Kernel Release Notes
    • 3.2.2.3. Known Issues
  • 3.2.3. RT Kernel Release Notes
    • 3.2.3.1. Build Information
    • 3.2.3.2. Generic Kernel Release Notes
    • 3.2.3.3. Known Issues
  • 3.2.4. Kernel Drivers
    • 3.2.4.1. ADC
    • 3.2.4.2. Audio
      • 3.2.4.2.1. Introduction
      • 3.2.4.2.2. Software Architecture
      • 3.2.4.2.3. Generic commands and instructions
      • 3.2.4.2.4. Board-specific instructions
      • 3.2.4.2.5. Potential issues
      • 3.2.4.2.6. Additional Information
    • 3.2.4.3. Crypto
    • 3.2.4.4. DCAN
    • 3.2.4.5. LCDC
    • 3.2.4.6. GPIO
    • 3.2.4.7. I2C
    • 3.2.4.8. CPSW
      • 3.2.4.8.1. Introduction
      • 3.2.4.8.2. Common Platform Time Sync (CPTS) module
    • 3.2.4.9. PRUSS
    • 3.2.4.10. PWM
    • 3.2.4.11. SPI
    • 3.2.4.12. NAND
    • 3.2.4.13. MMC/SD
      • 3.2.4.13.1. Introduction
      • 3.2.4.13.2. References
      • 3.2.4.13.3. Acronyms & Definitions
      • 3.2.4.13.4. Features
      • 3.2.4.13.5. Supported High Speed Modes
      • 3.2.4.13.6. Driver Configuration
      • 3.2.4.13.7. Steps for working around SD card issues
    • 3.2.4.14. UART
      • 3.2.4.14.1. Introduction
      • 3.2.4.14.2. Driver Features
      • 3.2.4.14.3. Basic External loopback testing
    • 3.2.4.15. UFS
    • 3.2.4.16. USB MUSB
  • 3.2.5. LTP-DDT Validation
  • 3.2.6. FAQs
• 3.3. Filesystem
• 3.4. Tools
  • 3.4.1. Development Tools
    • 3.4.1.1. Processor SDK Linux Top-Level Makefile
    • 3.4.1.2. Processor SDK Linux GCC Toolchain
    • 3.4.1.3. Creating SD Cards
    • 3.4.1.4. Processor SDK Linux Setup Script
  • 3.4.2. Flash Tools
    • 3.4.2.1. AM335x/AM437x Flash Programming
    • 3.4.2.2. Sitara Uniflash
  • 3.4.3. Pin Mux Tools
  • 3.4.4. Code Composer Studio
    • 3.4.4.1. CCS Installation
    • 3.4.4.2. CCS Compiling
    • 3.4.4.3. Remote Explorer Setup with CCS
    • 3.4.4.4. GDB Setup with CCS
    • 3.4.4.5. Kernel Debugging with CCS
• 3.5. PRU Subsystem
  • 3.5.1. Overview of PRU Subsystem
    • 3.5.1.1. Getting Started Information
    • 3.5.1.2. Hardware Information
      • 3.5.1.2.1. Technical Documentation
      • 3.5.1.2.2. PRU Differences Between Devices
      • 3.5.1.2.3. Miscellaneous
    • 3.5.1.3. Training Material
    • 3.5.1.4. Development Tools
      • 3.5.1.4.1. PRU C Compiler
      • 3.5.1.4.2. PRU Assembly Instructions
    • 3.5.1.5. Software Information
      • 3.5.1.5.1. Firmware Software Information
      • 3.5.1.5.2. U-Boot Software Information
    • 3.5.1.6. Examples
    • 3.5.1.7. Evaluation Hardware
    • 3.5.1.8. TI Designs
    • 3.5.1.9. Support
    • 3.5.1.10. PRU FAQ
      • 3.5.1.10.1. PRU Applications & Support questions
        • 3.5.1.10.1.1. What is the difference between the PRU subsystem and ICSS?
        • 3.5.1.10.1.2. Is TI providing libraries for the PRU?
        • 3.5.1.10.1.3. Can I develop my own industrial protocols on the PRU-ICSS?
        • 3.5.1.10.1.4. Can the PRU run a High Level Operating System?
        • 3.5.1.10.1.5. My processor has a PRU. Is the PRU supported in the Linux Processor SDK?
      • 3.5.1.10.2. PRU Memory Access questions
        • 3.5.1.10.2.1. Why does my PRU firmware hang when reading or writing to an address external to the PRU Subsystem?
        • 3.5.1.10.2.2. In AM437x, why can PRU-ICSS0 not access memories outside of the ARM?
        • 3.5.1.10.2.3. Why can the PRU not edit pinmux settings?
      • 3.5.1.10.3. PRU GPI/O questions
        • 3.5.1.10.3.1. What is the maximum speed for toggling PRU GPO pins via PRU software?
        • 3.5.1.10.3.2. When does the PRU start capturing from the input pins?
        • 3.5.1.10.3.3. Can the module be modified so that the GPI start bit is a zero instead of a one?
        • 3.5.1.10.3.4. What happens after 28 bit GPI shifts?
        • 3.5.1.10.3.5. Can data be pre-loaded into shadow registers prior to configuring the PRU GPO mode to shift out mode?
        • 3.5.1.10.3.6. When does PRU<n>_CLOCKOUT start running?
        • 3.5.1.10.3.7. When does the PRU start shifting data in the shadow registers?
        • 3.5.1.10.3.8. The shadow registers are loaded by writing to PRU<n>_R30 [0:15]. Does this change the state of the corresponding device-level pins?
        • 3.5.1.10.3.9. When the PRU<n>_ENABLE_SHIFT bit is cleared, does the PRU immediately stop shifting PRU<n>_DATAOUT?
        • 3.5.1.10.3.10. Does the PRU shift data out LSB or MSB first?
        • 3.5.1.10.3.11. What happens to the content stored in R30 when the PRU changes to a different GPO mode?
      • 3.5.1.10.4. PRU INTC and System Event questions
        • 3.5.1.10.4.1. How can a PRU core interrupt the ARM? How can the ARM interrupt a PRU core?
        • 3.5.1.10.4.2. On devices with multiple PRU-ICSSs, how can one PRU-ICSS interrupt the other?
      • 3.5.1.10.5. PRU Debugger questions
        • 3.5.1.10.5.1. When using the XDS510 USB emulator, why does the PRU Program Counter not increment correctly when stepping through PRU Disassembly code?
        • 3.5.1.10.5.2. Why are no MMRs outside the PRU subsystem visible from the PRU perspective memory browser window in CCS?
  • 3.5.2. Training
    • 3.5.2.1. PRU Getting Started Labs
      • 3.5.2.1.1. Introduction
        • 3.5.2.1.1.1. Supported Processors
        • 3.5.2.1.1.2. Required Hardware and Software
        • 3.5.2.1.1.3. Last Tested Configuration
    • 3.5.2.2. Lab 1: How to Create a PRU Project
      • 3.5.2.2.1. Creating a CCS PRU Project
        • 3.5.2.2.1.1. Install and Set up CCS
        • 3.5.2.2.1.2. Create the PRU Project
      • 3.5.2.2.2. Creating a Linux PRU Project
    • 3.5.2.3. Lab 2: How to Write PRU Firmware
      • 3.5.2.3.1. Writing C Code
        • 3.5.2.3.1.1. Writing C Code in CCS
        • 3.5.2.3.1.2. Writing C Code in Linux
      • 3.5.2.3.2. Writing Assembly Code
        • 3.5.2.3.2.1. Writing Assembly Code in CCS
        • 3.5.2.3.2.2. Writing Assembly Code in Linux
      • 3.5.2.3.3. Writing Mixed C and Assembly Code
        • 3.5.2.3.3.1. Write the Main Function in C
        • 3.5.2.3.3.2. Write the External Function in Assembly
        • 3.5.2.3.3.3. More Resources for Writing Mixed C and Assembly Code
      • 3.5.2.3.4. Writing Mixed C and Inline Assembly Code
        • 3.5.2.3.4.1. Write the Main Function in C
        • 3.5.2.3.4.2. Write the Inline Function in Assembly
        • 3.5.2.3.4.3. More Resources for Writing Mixed C and Inline Assembly Code
    • 3.5.2.4. Lab 3: How to Compile PRU Firmware
      • 3.5.2.4.1. Compiling From CCS
        • 3.5.2.4.1.1. Compiling to a .out File
        • 3.5.2.4.1.2. Compiling to a Hex Array File
        • 3.5.2.4.1.3. CCS Compiler Settings
      • 3.5.2.4.2. Compiling From Linux
        • 3.5.2.4.2.1. Linux Compiler Settings
    • 3.5.2.5. Lab 4: How to Initialize the PRU
      • 3.5.2.5.1. Initializing the PRU from CCS
        • 3.5.2.5.1.1. Set up the Target Configuration
        • 3.5.2.5.1.2. Connect to the EVM
        • 3.5.2.5.1.3. Test the Target Configuration
        • 3.5.2.5.1.4. Load the PRU firmware
      • 3.5.2.5.2. Initializing the PRU from Linux Core
        • 3.5.2.5.2.1. Configure the Kernel
        • 3.5.2.5.2.2. Modify Device Tree Files to account for Pinmuxing
        • 3.5.2.5.2.3. Copy files to the target filesystem
        • 3.5.2.5.2.4. Boot the new device tree and FS on the Target
    • 3.5.2.6. Lab 5: Basic Debugging of PRU Firmware
      • 3.5.2.6.1. Debugging the PRU from CCS
        • 3.5.2.6.1.1. Debugging C Code
        • 3.5.2.6.1.2. Debugging Assembly Code
      • 3.5.2.6.2. Debugging the PRU from Linux Core
    • 3.5.2.7. PRU Hands-on Labs
      • 3.5.2.7.1. Lab Configuration
        • 3.5.2.7.1.1. Hardware
        • 3.5.2.7.1.2. Software
        • 3.5.2.7.1.3. Supported Platforms
      • 3.5.2.7.2. LAB 1: Toggle LED with PRU GPO
        • 3.5.2.7.2.1. Create & Build the PRU Project
        • 3.5.2.7.2.2. Load the PRU Firmware
      • 3.5.2.7.3. LAB 2: Read Push Button Switch on PRU0 GPI & Toggle LED with PRU1 GPO
        • 3.5.2.7.3.1. Build the PRU Firmware
        • 3.5.2.7.3.2. Load the PRU Firmware
        • 3.5.2.7.3.3. Debug the PRU Firmware
      • 3.5.2.7.4. LAB 3: Temperature Monitor
        • 3.5.2.7.4.1. Design & Build the PRU Firmware
        • 3.5.2.7.4.2. Load the PRU Firmware
        • 3.5.2.7.4.3. Debug the PRU Firmware
      • 3.5.2.7.5. LAB 4: Introduction to Linux driver
        • 3.5.2.7.5.1. Build the PRU Firmware
        • 3.5.2.7.5.2. Configure the Kernel
        • 3.5.2.7.5.3. Build the Linux Kernel and remoteproc Driver
        • 3.5.2.7.5.4. Modify Device Tree Files to Account for PRU Cape
        • 3.5.2.7.5.5. Copy files to the target filesystem
        • 3.5.2.7.5.6. Boot the new device tree and FS on the Target
      • 3.5.2.7.6. LAB 5: RPMsg Communication between ARM and PRU
        • 3.5.2.7.6.1. Build the PRU Firmware - Using CCSv6
        • 3.5.2.7.6.2. Build the PRU Firmware - Using the Provided Makefile
        • 3.5.2.7.6.3. Build the RPMsg Client Sample Driver
        • 3.5.2.7.6.4. Copy files to the target file system
        • 3.5.2.7.6.5. Part 1: Kernel space communication
        • 3.5.2.7.6.6. What just happened?
        • 3.5.2.7.6.7. Part 2: User Space Communication
        • 3.5.2.7.6.8. Part 3: User Space Application
        • 3.5.2.7.6.9. What just happened?
      • 3.5.2.7.7. LAB 6: Blinking LEDs with RPMsg from Linux User Space
        • 3.5.2.7.7.1. Build the PRU Firmware - Using CCSv6
        • 3.5.2.7.7.2. Build the PRU Firmware - Using the Provided Makefile
        • 3.5.2.7.7.3. Copy files to the target file system
        • 3.5.2.7.7.4. Part 1: Linux Command Line LED Toggling
        • 3.5.2.7.7.5. What just happened?
        • 3.5.2.7.7.6. Part 2: Linux Shell Script LED Toggling
        • 3.5.2.7.7.7. What just happened?
    • 3.5.2.8. Getting Started with PRU Software Support Package
    • 3.5.2.9. RPMsg Quick Start Guide
      • 3.5.2.9.1. Introduction
        • 3.5.2.9.1.1. Supported Devices
        • 3.5.2.9.1.2. Hardware Assumptions
      • 3.5.2.9.2. Getting the Linux Processor SDK
      • 3.5.2.9.3. Configuring and Building the Linux Kernel with RPMsg Support
      • 3.5.2.9.4. Creating a Bootable SD Card with RPMsg Support
      • 3.5.2.9.5. Booting the Board and Testing RPMsg
        • 3.5.2.9.5.1. Out of Box Demo Explanation
      • 3.5.2.9.6. Getting Started with RPMsg Development
        • 3.5.2.9.6.1. Rebuilding the PRU Firmwares from Source
        • 3.5.2.9.6.2. Placing the Rebuilt Firmware into the Embedded Linux File System
        • 3.5.2.9.6.3. Loading the New Code into the PRUs and Running
      • 3.5.2.9.7. Common Issues
        • 3.5.2.9.7.1. Board is Not Set Up for SD Card Boot
  • 3.5.3. Linux Drivers
    • 3.5.3.1. RemoteProc
    • 3.5.3.2. RPMsg
      • 3.5.3.2.1. Introduction
        • 3.5.3.2.1.1. Supported Devices
    • 3.5.3.3. PRU-ICSS Soft UART
    • 3.5.3.4. PRU-ICSS Serial UART
  • 3.5.4. Firmware Development
    • 3.5.4.1. Resource Tables
      • 3.5.4.1.1. When are resource tables used?
      • 3.5.4.1.2. Linux kernel 5.10 or later
      • 3.5.4.1.3. Linux kernel 5.4 or earlier
    • 3.5.4.2. INTC Configuration
      • 3.5.4.2.1. Linux kernel 5.10 or later
      • 3.5.4.2.2. Linux kernel 5.4 or earlier
    • 3.5.4.3. Header Files
  • 3.5.5. Hardware
    • 3.5.5.1. PRU EVMs
    • 3.5.5.2. PRU Cape Hardware User Guide
      • 3.5.5.2.1. Introduction
        • 3.5.5.2.1.1. Description
        • 3.5.5.2.1.2. EVM System View
      • 3.5.5.2.2. Schematics/Design/Errata Files
      • 3.5.5.2.3. System Description
        • 3.5.5.2.3.1. System Board Diagram
        • 3.5.5.2.3.2. Signals Used
      • 3.5.5.2.4. PRU Cape Functional Block Descriptions
        • 3.5.5.2.4.1. Audio
        • 3.5.5.2.4.2. EEPROM
        • 3.5.5.2.4.3. eCAP
        • 3.5.5.2.4.4. LCD
        • 3.5.5.2.4.5. LEDs
        • 3.5.5.2.4.6. Switches
        • 3.5.5.2.4.7. Temperature Sensor
        • 3.5.5.2.4.8. Test Space
        • 3.5.5.2.4.9. UART
        • 3.5.5.2.4.10. Legal
    • 3.5.5.3. PRU Cape Getting Started Guide
    • 3.5.5.4. PRU Cape Building Demos
• 3.6. Graphics and Display
  • 3.6.1. Introduction
  • 3.6.2. Software Architecture
  • 3.6.3. Graphics Demos
    • 3.6.3.1. Running demos in Null Window system mode
  • 3.6.4. SGX Debug Info
    • 3.6.4.1. Introduction
    • 3.6.4.2. Baselining the current SGX driver environment
    • 3.6.4.3. Run-time checks/configuration of the SGX driver
      • 3.6.4.3.1. WindowSystem
      • 3.6.4.3.2. DisableHWTextureUpload
      • 3.6.4.3.3. DefaultPixelFormat
    • 3.6.4.4. SGX Driver Failure Modes (Installation)
      • 3.6.4.4.1. Unable to install the kernel modules (pvrsrvkm.ko)
    • 3.6.4.5. SGX Driver Failure Modes (Run time)
      • 3.6.4.5.1. Vertical Tearing / Artifacts / Clipping issues / Missing objects
      • 3.6.4.5.2. Demos are not running at required speed, How to check SGX clock rate?
      • 3.6.4.5.3. Qt demos do not work when PowerVR is enabled
    • 3.6.4.6. Posting to E2E forum
  • 3.6.5. AM3 Beagle Bone Black Board Configuration
  • 3.6.6. Migration from prior releases
    • 3.6.6.1. From Processor SDK 1.x to 2.x for AM3, AM4
      • 3.6.6.1.1. Window System Libraries
      • 3.6.6.1.2. Obsolete Test Programs
    • 3.6.6.2. From Processor SDK 2.0.0 to 2.0.x for AM4
      • 3.6.6.2.1. Window System Libraries
    • 3.6.6.3. From Processor SDK 2.0.1 to 2.0.x for AM3/4/5
      • 3.6.6.3.1. Window System Libraries
    • 3.6.6.4. From Processor SDK 3.1 to 3.x for AM3/4/5
    • 3.6.6.5. From Processor SDK 6.1 to 6.2 for AM3/4/5/6
      • 3.6.6.5.1. Window System Libraries
  • 3.6.7. DSS Applications
    • 3.6.7.1. Running DSS Applications
      • 3.6.7.1.1. Running drmclone
      • 3.6.7.1.2. Running drmextended
      • 3.6.7.1.3. Running drmzalpha
  • 3.6.8. SGX Build Guide
    • 3.6.8.1. SGX Kernel Module
      • 3.6.8.1.1. Building the SGX Kernel Module
    • 3.6.8.2. SGX Userspace Components
      • 3.6.8.2.1. Installing the SGX Userspace Components
    • 3.6.8.3. SGX Mesa Components
      • 3.6.8.3.1. Building the SGX Mesa Components
    • 3.6.8.4. Using the SGX Stack
  • 3.6.9. Display
    • 3.6.9.1. Finding Connector ID
    • 3.6.9.2. Finding Plane ID
    • 3.6.9.3. Using Connector ID and Plane ID
  • 3.6.10. OpenGL ES
  • 3.6.11. QT Graphics Framework
    • 3.6.11.1. Demos
    • 3.6.11.2. QT QPA
    • 3.6.11.3. Running QT Applications with the Weston IVI shell
  • 3.6.12. GTK+ Graphics Framework
    • 3.6.12.1. Demos
  • 3.6.13. Weston
    • 3.6.13.1. Starting Weston with Systemd
    • 3.6.13.2. Starting Weston Manually
    • 3.6.13.3. Stopping Weston
    • 3.6.13.4. Running Weston clients
    • 3.6.13.5. Running multimedia with Wayland sink
    • 3.6.13.6. Using IVI shell feature
  • 3.6.14. PowerVR Tools
    • 3.6.14.1. PVRTune
    • 3.6.14.2. PVRCarbon
• 3.7. OpenCV
  • 3.7.1. Introduction
  • 3.7.2. OpenCV Modules Supported By TI
  • 3.7.3. OpenCL offload
  • 3.7.4. Unit Tests
  • 3.7.5. Necessary steps to modify OpenCV framework to add more OpenCL Host side and DSP C66 optimized kernels
    • 3.7.5.1. Supported Platforms
    • 3.7.5.2. OpenCV OpenCL run-time setup
    • 3.7.5.3. OpenCV OpenCL development setup
    • 3.7.5.4. OpenCV OpenCL related framework details: how to add new DSP kernel
    • 3.7.5.5. Creating OpenCL C kernel optimized for C66 core
    • 3.7.5.6. OpenCV OpenCL kernels implemented specifically for DSP C66 core
      • 3.7.5.6.1. EDMA transfer framework
      • 3.7.5.6.2. Additional information about C66 specific optimizations
    • 3.7.5.7. List of currently (SDK 3.1) DSP optimized OpenCV OpenCL kernels, using non-standard OpenCL extensions
      • 3.7.5.7.1. OpenCL C C66 DSP kernels
    • 3.7.5.8. Profiling results of DSP optimized OpenCV OpenCL kernels (SDK 3.1), AM5728 platform
      • 3.7.5.8.1. Single channel, 1200x709, barcode ROI detection use case
      • 3.7.5.8.2. Single channel, 1920x1080. barcode ROI detection use case
      • 3.7.5.8.3. Single channel, 720x576, Gesture recognition use case
    • 3.7.5.9. Alternative approach to add new OpenCL kernels at OpenCV application level
      • 3.7.5.9.1. OpenCL kernel dispatch from OpenCV application, using existing OpenCV-OpenCL classes
      • 3.7.5.9.2. OpenCL kernel dispatch from OpenCV application, using standard OpenCL dispatch with access to OpenCV data objects
    • 3.7.5.10. OpenCV profiling - standard procedure
  • 3.7.6. OpenCV Performance
  • 3.7.7. Frequently Asked Questions
• 3.8. Virtualization
  • 3.8.1. Docker Linux Container Runtime
    • 3.8.1.1. Overview
    • 3.8.1.2. Exploring Docker In Processor SDK
      • 3.8.1.2.1. Booting the Docker Filesystem
      • 3.8.1.2.2. Configuring the Docker daemon
      • 3.8.1.2.3. Docker Hello World
    • 3.8.1.3. Docker References
• 3.9. Machine Learning
  • 3.9.1. TI Deep Learning (TIDL)
    • 3.9.1.1. Introduction
      • 3.9.1.1.1. Deep Learning Inference in Embedded Device
      • 3.9.1.1.2. Application space
      • 3.9.1.1.3. Supported Devices
      • 3.9.1.1.4. TIDL API framework abstracts multicore operation
    • 3.9.1.2. Verified networks topologies
      • 3.9.1.2.1. Neural network layers supported by TIDL
      • 3.9.1.2.2. Constraints on layer parameters
    • 3.9.1.3. Examples and Demos
      • 3.9.1.3.1. TIDL API examples
      • 3.9.1.3.2. Matrix GUI demos
        • 3.9.1.3.2.1. Description of Matrix-GUI classification example
        • 3.9.1.3.2.2. Description of Matrix-GUI object detection example
        • 3.9.1.3.2.3. Description of Matrix-GUI image segmentation example
    • 3.9.1.4. Developer’s guide
      • 3.9.1.4.1. Software Stack
      • 3.9.1.4.2. Additional public TI resources
      • 3.9.1.4.3. Introduction to Programming Model
      • 3.9.1.4.4. Target file-system
        • 3.9.1.4.4.1. Firmware
        • 3.9.1.4.4.2. User space components
      • 3.9.1.4.5. Input data format
      • 3.9.1.4.6. Output data format
      • 3.9.1.4.7. Import Process
        • 3.9.1.4.7.1. Configuration Parameters for Import
        • 3.9.1.4.7.2. Sample configuration file
        • 3.9.1.4.7.3. Import tool traces
        • 3.9.1.4.7.4. Splitting layers between layers groups
        • 3.9.1.4.7.5. Calculating theoretical GMACs needed
        • 3.9.1.4.7.6. Mapping to EVE capabilities
      • 3.9.1.4.8. Verifying TIDL inference result
      • 3.9.1.4.9. Parameters controling dynamic quantization
      • 3.9.1.4.10. Importing Tensorflow Models
      • 3.9.1.4.11. Importing Caffe Models
      • 3.9.1.4.12. Viewer tool
      • 3.9.1.4.13. Simulation Tool
      • 3.9.1.4.14. Summary of model porting steps
    • 3.9.1.5. Compatibility of trained model formats
    • 3.9.1.6. Training
      • 3.9.1.6.1. Example of training procedure
      • 3.9.1.6.2. Where does the benefit of sparsification come from?
    • 3.9.1.7. Performance data
      • 3.9.1.7.1. Computation performance of verified networks
    • 3.9.1.8. Multi core performance (EVE and DSP cores only)
      • 3.9.1.8.1. Accuracy of selected networks
    • 3.9.1.9. Troubleshooting
  • 3.9.2. Neo-AI Deep Learning Runtime
  • 3.9.3. TVM Runtime
    • 3.9.3.1. Introduction
    • 3.9.3.2. Build and Install TVM Compiler
    • 3.9.3.3. Running Inference on the Target
  • 3.9.4. TensorFlow Lite
    • 3.9.4.1. Introduction
    • 3.9.4.2. Supported version
    • 3.9.4.3. TensorFlow Lite example applications
      • 3.9.4.3.1. Running benchmark_model
      • 3.9.4.3.2. Running label_image example
    • 3.9.4.4. TensorFlow Lite classification and segmentation demos with OpenCV
      • 3.9.4.4.1. Running classification demo
      • 3.9.4.4.2. Running segmentation demo
      • 3.9.4.4.3. Using video clip and camera capture as input
      • 3.9.4.4.4. Rebuild the Tensorflow Lite demos
    • 3.9.4.5. Tensorflow Lite heterogeneous execution with TIDL compute offload
      • 3.9.4.5.1. Offline TIDL compilation of Tensorflow Lite mode
      • 3.9.4.5.2. Tensorflow Lite runtime execution with TIDL offload
      • 3.9.4.5.3. Helper scripts for out of box experience
      • 3.9.4.5.4. Rebuild Tensorflow Lite demos with TIDL offload
      • 3.9.4.5.5. Limitations
  • 3.9.5. Arm NN and Arm Compute Library
    • 3.9.5.1. Introduction
    • 3.9.5.2. Supported versions
    • 3.9.5.3. Arm Compute Library
    • 3.9.5.4. Arm NN
    • 3.9.5.5. Arm NN MobileNet Demo