Renesas RZ/V2H Cortex-A55/R8/M33 MPU comes with 80 TOPS AI accelerator for robotics and autonomous applications

The RZ/V2H is the latest addition to the Reneasas family of 32-bit and 64-bit microprocessors. Renesas RZ/V2H microprocessor features four Arm Cortex-A55 application processing cores, dual Arm Cortex-R8 real-time processing cores, and one Cortex-M33 core for system management, as well as an AI accelerator with up to 80 TOPS (sparse) of performance.

Renesas RZ/V2H block chart

The RZ/V2H MPU includes the third generation of Renesas’ dynamically reconfigurable processor for AI (DRP-AI3) accelerator, a staple of the RZ/V series. The DRP-AI accelerator can run complex image AI models at a power efficiency of 10 TOPS per watt (TOPS/W), as much as 10 times higher than conventional microprocessors. Additionally, the RZ/V2H has another DRP that can be used in image processing and robotics applications.

The board also comes with PCIe Gen3, USB 3.2, and Gigabit Ethernet for high-speed applications such as autonomous robotics and factory automation machine vision. It offers a performance boost over previous products such as the RZ/G2L MPU, which allows for processing more data in real-time and reduces the need for cloud computing platforms.

Renesas RZ/V series ai inference performance

Renesas RZ/V2H specifications:

  • CPU
    • 4x 64-bit Arm Cortex-A55 cores for application processing (up to 1.8 GHz)
    • 2x 32-bit Arm Cortex-R8 cores for real-time processing (up to 800 MHz)
    • 1x 32-bit Arm Cortex-M33 core for system management (up to 200 MHz)
  • Memory
    • 6MB on-chip SRAM with ECC
    • External DDR memory interface
    • 2-channel memory controller for LPDDR4-3200 or LPDDR4X-3200 with a 32-bit bus width
    • xSPI interface
    • SDHI (eMMC/SD)
  • AI accelerator engines
    • AI accelerator (dynamically reconfigurable processor for AI (DRP-AI3) – 8 TOPS/dense, 80 TOPS/sparse
    • Dynamically reconfigurable processor (DRP)
  • Video & Graphics
    • GPU – Arm Mali-G31 (optional)
    • Mali C-55 image signal processor (ISP) (optional)
    • Image scaling unit (ISU)
    • Video codec unit (VCD)
  • Display – MIPI DSI
  • Camera – MIPI CSI-2
  • Network – 2x Gigabit Ethernet
  • USB – 1x USB 2.0 (Host, OTG), 1x USB 3.2 Gen2 (2ch: Host-only)
  • PCIe – PCIe Gen3
  • Other Peripherals
    • 6x CAN/CAN FD (compliant with ISO11898-1)
    • 3x SPI, 9x I2C, 1x I3C
    • 8x 12-bit ADC
  • Supply Voltage –  3.135V – 3.465V
  • Package – BGA, 19mm x 19mm, 1368-pin, 0.5mm pitch
  • Temperature Range – -40 to 125°C

Renesas has partnered with SEGGER to provide support for the RZ/V2H via the J-link debug probe. The company has also released a library of pre-trained models and an AI Linux software development kit (SDK) to aid rapid development. There is an evaluation board, RZ/V2H-EVK, that can be used to test AI applications early in the design phase.

Renesas RZ/V2H Evaluation Kit

The V2H microprocessor and its evaluation kit are available now, with the latter priced at $1,176. Pricing for the RZ/V2H can be found on the product page. You can request free samples on the website and learn more about the microprocessor and its development tools by checking out the product page and press release.

Thanks to TLS for the tip.

Share this:

Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress

ROCK Pi 4C Plus
Subscribe
Notify of
guest
The comment form collects your name, email and content to allow us keep track of the comments placed on the website. Please read and accept our website Terms and Privacy Policy to post a comment.
1 Comment
oldest
newest
1 month ago

If that ResNet50 chart is to be believed, that is closer to 25 dense TOPS in direct comparison, or there is an overall limiting factor at play.

Khadas VIM4 SBC