MediaTek i350 Edge AI Processor is Designed for Mainstream AIoT & Smart Home Applications

MediaTek is well known for its mobile processors, but recently the company also manufactures IoT and AI processors such as MediaTek i300/i500 and MediaTek i700.

The company has now introduced an updated to the entry-level MediaTek i300 processor with MediaTek i350 quad-core Cortex-A53 processor integrated with DSP and AI processor for vision and voice edge processing applications including facial, object, gesture, motion recognition, license plate recognition (LPR), voice activation and speed recognition, sound isolation, bio-tech and biometric measurements, and more.

MediaTek i350

MediaTek i350 key features and specifications:

  • Processor – 4x Arm Cortex-A53 cores up to 2.0 GHz with 512KB unified L2 cache, Arm NEON
  • GPU – Arm Mali-G52 MC1 up to 800Mhz with support for OpenGL ES, OpenCL, and Vulkan APIs
  • Co-processor – MediaTek APU 1.0 AI Processor up to 500MHz (285 GMACs)
  • Memory -LPDDR4/LPDDR4X up to 3200Mbps, LPDDR3, DDR3/L up to 1866Mbps
  • Storage – eMMC 5.1 flash (SLC/MLC/TLC)
  • Display Interfaces
    • LVDS, HDMI, MIPI DSI, MediaTek DPI with 8-bit RGB/YUV444, or 8/10/10-bit YUV422 pixel data on a parallel output bus.
    • Dual display configuration up to 2x 1920×1200
  • Camera
    • 13MP ISP up to 1920×1080 resolution, 1080p6 H.265/H.264 video recording
    • MIPI CSI interface
  • Video
    • Encoding – H.264, H.265 / HEVC
    • Decoding: H.264, H.265 / HEVC, MPEG-1/2/4, VP9
  • Audio – Audio Amplifier, HiFi4 DSP, 8-channel PDM-in, 8-channel TDM-in
  • Networking and Wireless Connectivity
    • 10/100M Fast Ethernet
    • Dual-band 802.11b/g/n/ac WiFi 5 in 1T1R or 2T2T antenna configuration and  Bluetooth 5.0 via the unannounced MT7663 chipset
    • Optional GNSS and FM radio support
  • USB – USB 2.0 OTG and host
  • Other Peripherals – GPIO, I2C, SDIO
  • Process – 14nm

The product page does not say anything about software support, but I’d assume both Linux and Android BSP’s are available, and MediaTek i350 is compatible with Neuropilot SDK, but sadly the documentation requires a username/password.

The illustration provided by MediaTek indicated the i350 Edge AI processor will mostly target smart home products such as smart washing machines, smart dishwashers, vacuum robots, etc… Further information may be found on the product page.

Share this:
FacebookTwitterHacker NewsSlashdotRedditLinkedInPinterestFlipboardMeWeLineEmailShare

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

ROCK Pi 4C Plus

3 Replies to “MediaTek i350 Edge AI Processor is Designed for Mainstream AIoT & Smart Home Applications”

    1. Good question. Apparently MP represents the number of pixels per clock, and MC the number of physical cores. Some GPUs have shaders of different sizes with for examples a 2-shader GPU with a 2-pixels-per-clock shader and a 1-pixel-per-clock shaders for a total of 3-pixels per clock. So If I understand correctly, this example would correspond to MP3 or MC2.

      Source: https://community.arm.com/developer/tools-software/graphics/f/discussions/46737/i-want-to-know-about-the-gpu-line-up

  1. SDK for this is supposed to be open source from Bay Libre, but I can’t locate it.

    I also wonder how much this chipset costs since the i300 dev boards are $200.

Leave a Reply

Your email address will not be published. Required fields are marked *

Khadas VIM4 SBC
Khadas VIM4 SBC