Vicharak Axon SBC features RK3588, 8GB RAM, 4K Graphics, and 6 TOPS NPU for AI Applications

Vicharak recently introduced the Axon—AI Edge Single Board Computer (SBC), a compact yet powerful computing device designed for AI, graphics, and multi-display workflows.

Built around the RockChip RK3588 the Axon platform is a high-performance computing system powered by the RockChip RK3588 processor, featuring an 8-core CPU (4x Cortex-A76 + 4x Cortex-A55) clocked up to 2.4GHz and an ARM Mali-G610 MP4 GPU with 450 GFLOPS. Designed for AI and machine learning workloads, it includes a 6-TOPS NPU supporting INT4, INT8, and INT16 operations, compatible with TensorFlow, MXNet, PyTorch, and Caffe.

The platform offers memory configurations ranging from 4GB to 32GB LPDDR5 and storage options up to 128GB eMMC, expandable via M.2 (PCIe 2.0 NVMe SSD) or MicroSD. It supports 8K video decoding/encoding, HDMI 2.1, MIPI-DSI, and DP1.4 for multi-screen setups. Connectivity includes dual-band WiFi 6, Bluetooth 5.2, Gigabit Ethernet, USB 3.0, USB-C, and MIPI-CSI camera interfaces. Compatible with Android 12 and Ubuntu 22.04, it is used in AI model processing, robotics, edge computing, and multimedia applications.

Previously, Vicharak developed the Vaaman SBC, designed for advanced computing tasks with a six-core ARM CPU and an Efinix Trion T120 FPGA with 112,128 logic elements. We have also covered other SBCs with AI features including the BeagleBoard’s New Beagley-AI, the Axelera Metis, the iWave System’s iW-RainboW-G54S, and others. Feel free to check those out, if you are interested in this topic.

Axon AI SBC Specifications:

• SoC: Rockchip RK3588
  • CPU:
    • 8-core 64-bit (4x Cortex-A76 + 4x Cortex-A55), up to 2.4GHz
  • GPU:
    • ARM Mali-G610 MP4 quad-core GPU, 450 GFLOPS
    • Supports OpenGL ES3.2, OpenCL 2.2, Vulkan 1.1
  • NPU:
    • 6 TOPS for AI acceleration
    • Supports INT4, INT8, INT16 mixed operations
    • Compatible w/ TensorFlow, MXNet, PyTorch, Caffe
  • ISP: 48MP ISP w/ HDR and DNR
  • VPU:
    • Decoding:
      • 8K@60fps (H.265, VP9, AVS2)
      • 8K@30fps (H.264 AVC/MVC)
      • 4K@60fps (AV1)
      • 1080P@60fps (MPEG-2/-1/VC-1/VP8)
    • Encoding:
      • 8K@30fps (H.265/H.264)
      • Up to 32-channel 1080P@30fps decoding
      • 16-channel 1080P@30fps encoding
• Memory:
  • 4GB to 16GB LPDDR4x
  • Expandable to 32GB LPDDR5
• Storage:
  • 32GB to 128GB eMMC
  • 1x eMMC 5.1
  • 2x PCIe 2.0
  • 4x PCIe 3.0(NVMe M.2)
  • Micro SD Card slot (up to 32GB)
  • SATA 3.0
• Display:
  • 2x HDMI TX 2.1
  • 2x MIPI DSI 2.0
  • 1x Type-C DP 1.4a
  • Supports up to 4x displays
• Audio:
  • 3.5mm Audio Jack
  • Speakers
  • Analog Mic
  • HDMI 2.1 audio output
  • DP1.4 audio output
• Video:
  • 2x HDMI RX 2.0
  • 2x MIPI Camera Interface 1.2/2.0
  • 4K video playback and multi-display support
• Connectivity:
  • 1x 1000Mbps Ethernet (RJ45)
  • 2.4GHz and 5GHz dual-band WiFi 6
  • Bluetooth 5.2
• USB:
  • 2x USB 2.0
  • 2x USB 3.1
  • 1x USB-C multifunction interface (USB3.0 OTG/DP1.4)
• I/O:
  • 4x I2C
  • 1x SPI
  • 1x I2S
  • 4x UART
  • 2x PCIe 2.0
  • 3x PCIe 3.0
• Expansion:
  • 30 pin GPIO
  • 1x Micro SD Card slot
• Misc:
  • 2x PWM
  • SARADC
• Power: 12V DC via PD charger
• Dimensions: 100 x 72mm
• Temperature:
  • Standard range: 0°C to 60°C (commercial grade)
  • Extended range (if available): -20°C to 80°C (industrial grade)
• Weight: 200 g

Additionally, in terms of software support, this AI Edge Single Board Computer is compatible with Android 12 and Ubuntu 22.04, providing flexibility for various applications.

The Axon GitHub repository provides step-by-step instructions for setting up the platform’s NPU to run AI, ML, and DL models efficiently. It includes detailed guides on converting CNN models to RKNN format and adapting LLMs from Hugging Face or GGUF formats to RKLLM.

At the time of writing, the Axon SBC with 8GB RAM + 32GB eMMC is available via Vicharak on Amazon India for ₹18,290 (~$212.79). For more information, you may visit the documentation and product page.

Please follow and like us: