All News | Boards | Chips | Devices | Software | Archive | About | Contact | Subscribe
Follow LinuxGizmos:
Twitter Facebook Pinterest RSS feed
*   get email updates   *

QCS6490 Vision-AI Development Kit: Featuring 13 TOPs NPU and 8-Core Kryo 670 CPU

May 16, 2024 — by Giorgio Mendoza 238 views

Avnet has introduced the QCS6490 Vision-AI Development Kit, a sophisticated solution designed for vision-based AI applications. This kit includes an energy-efficient, multi-camera SMARC 2.1.1 compute module powered by the Qualcomm QCS6490 SoC.

The QCS6490 SOC, central to the SM2S-QCS6490 SMARC compute module, features an 8-core Kryo 670 CPU—comprising 4x Arm Cortex-A78 CPUs (up to 2.7 GHz) and 4x Arm Cortex-A55 CPUs (up to 1.9 GHz)—and an Adreno 643 GPU (up to 812 MHz).


It also includes a Hexagon DSP and a 6th generation AI Engine with 13 TOPS, alongside the Spectra 570L ISP for up to 64MP/30fps imaging and the Adreno 633 VPU for 4K video encoding and decoding.

QCS6490 Vision block diagram
(click image to enlarge)

The module includes 8GB LPDDR5 SDRAM and 64GB UFS Flash Memory, along with multiple USB, PCIe, and display interfaces, and two Gigabit Ethernet ports. Wi-Fi 6 and Bluetooth 5 are available as SMARC module assembly options, along with a TPM module. The module supports UART, SPI, I2C, I2S, and GPIO interfaces, as well as secure boot and hardware key management.

The Vision-AI IO Board, a core component of the QCS6490 Vision-AI Development Kit, supports a wide array of interfaces, enhancing connectivity and functionality. It includes a SMARC 2.1.1 edge connector and offers expansion options via M.2 key-M and key-E slots for NVME storage and Wi-Fi 6E/BT 5.4 respectively.

QCS6490 Vision top & bottom view
(click image to enlarge)

Connectivity features include a Gigabit Ethernet RJ45 connector, a variety of USB ports and display outputs through a miniDP and a MIPI-DSI connector. The board supports dual MIPI-CSI connectors for camera integration, and features comprehensive audio capabilities with two PDM microphones, an audio codec, and a stereo headphone jack.

The development kit is supplied with a Yocto-based Linux BSP and includes example AI and multi-camera applications. Support for Windows 11 IoT Enterprise is anticipated in the second half of 2024. Qualcomm will also offer support for Android and Ubuntu Linux on the QCS6490.

QCS6490 Vision dev kit
(click image to enlarge)

Specifications listed for the Vision-AI IO Board include:

  • Display/Audio:
    • 1x miniDP
    • 1x MIPI-DSI
    • 2x PDM Mics, Audio Codec, Stereo HP jack
  • Camera:
    • 1x 4L & 1x 2L MIPI-CSI 22-pin connectors
  • Connectivity:
    • 1x Gigabit Ethernet RJ45
  • Expansion:
    • M.2 key-M slot (NVME storage optional)
    • M.2 key-E slot (Wi-Fi 6E/BT 5.4 optional)
  • I/O Interface:
    • 40-Pin Pi-HAT header
    • 4-pin CAN header
    • 10-pin Test IO header
  • USB:
    • 2x USB 3.1
    • 2x USB 2.0
    • 1x USB-C OTG
  • Other Features:
    • 2-pin RTC battery header
    • 2x button switches & RGB User LED
  • Power:
    •  9V/3A (USB-C PD)
  • Operating Temperature:
    • -40°C to 85°C
  • Mechanical:
    • 100mm x 79mm
    • SMARC 2.1.1 edge connector (314-pin)

Further information

Avnet has not disclosed the pricing details for this product. The product announcement is available here.

(advertise here)

Print Friendly, PDF & Email

Please comment here...