RISC-V Solution

A flagship RISC-V mobile application processor reference

The Mobile SoC is a flagship RISC-V mobile application processor reference built on a broad in-house IP stack. It brings together a big, mid and little RISC-V CPU cluster, an NPU for on-device AI, an in-house GPU, a camera ISP, display and video codecs, LPDDR5X memory and UFS storage, and full connectivity, with zero ARM licensing anywhere on the die. It is a reference design in development.

Architecture

RISC-V CPU Cluster (big, mid and little cores) ACCELERATORS INTERCONNECT MEDIA AND IO NPU on-device AI ISP and Camera MIPI CSI-2 Connectivity USB, PCIe WiFi, Bluetooth On-Chip Interconnect coherent fabric and DMA GPU in-house graphics Display and Video Codec MIPI DSI, AV1 Security (HSM) hardware security module Memory LPDDR5X Controller UFS Storage

Clean architectural view. Every block is WIOWIZ IP on our own RISC-V cores, and internal micro-architecture is not shown.

RISC-V Architecture

A heterogeneous RISC-V CPU cluster of high-performance, efficiency and low-power cores hosts Linux and Android-class software, with an NPU for on-device AI and a full in-house media and connectivity stack alongside it. The graphics, ISP, display, video codec, memory and connectivity blocks are all WIOWIZ IP, with zero ARM license anywhere on the die. The design is fully controllable from spec to sign-off with our in-house EDA.

Specifications

CategoryMobile application processor (reference)
CPUHeterogeneous RISC-V cluster (performance, efficiency, low-power)
AIOn-chip NPU
GraphicsIn-house GPU
CameraISP with MIPI CSI-2
DisplayDisplay engine, MIPI DSI
VideoH.264, HEVC, AV1
MemoryLPDDR5X controller
StorageUFS
ConnectivityUSB, PCIe, WiFi, Bluetooth (modem optional)
SecurityHardware security module
SoftwareLinux, ONNX runtime
IP ownershipRISC-V and WIOWIZ IP, zero ARM license
Design flowWIOWIZ in-house EDA
StatusReference design, in development

Key Features

Heterogeneous RISC-V CPU cluster

A big, mid and little RISC-V core cluster hosts Linux and Android-class software with scalable performance and efficiency.

On-device AI (NPU)

An on-chip NPU runs neural inference locally for camera, voice and assistant workloads.

In-house GPU and ISP

A WIOWIZ GPU drives graphics and UI, paired with an ISP for MIPI CSI-2 camera capture.

LPDDR5X and UFS

An LPDDR5X memory controller and UFS storage give the platform high bandwidth and fast flash.

Zero ARM license

Built on RISC-V and WIOWIZ IP, with no ARM license anywhere on the die.

Full software stack (Linux, ONNX)

A complete Linux and ONNX runtime stack supports Android-class applications and portable AI models.

Applications

Smartphones Tablets Consumer application processors On-device AI