The AI Inference Platform by SEMIFIVE brings a robust solution for developing custom AI inference accelerators. Designed using silicon-verified components on Samsung's 14nm process, this platform offers an optimal balance between high performance and power efficiency, catering to demanding AI applications. It includes advanced IP such as a Quad-Core U74 RISC-V CPU and LPDDR4 memory, supporting high-speed PCIe Gen4 interfaces.
This platform is ideal for data center accelerators and AI vision processing, providing a comprehensive setup with package design, implementation, software, and driver support. The AI Inference Platform excels in swiftly transforming critical IPs and specifications into fully functional SoCs, significantly lowering the risks associated with design, time, and costs. SEMIFIVE’s Max Efficiency and Max Flexibility engagement models further enhance customization and development ease, making this platform a versatile choice for high-stakes AI projects.
Its capabilities include the capacity to operate at 2GHz frequencies, enabled by a hierarchical design structure with multiple gate and instance counts. Target applications extend to real-time processing tasks such as image and video recognition as well as advanced driver-assistance systems (ADAS). By leveraging mass production proven technology, it ensures a reliable transition from prototype to final production, making it a sought-after solution for cutting-edge AI applications.
