SiFive and AMD Demo Google LLM on RISC-V Platform

By AI123

At the AI Infra Summit on September 15, 2026, SiFive and AMD jointly demonstrated a server system pairing a RISC-V processor with an AMD Radeon AI PRO R9700 GPU, running Google's Gemma4-E2B model using the ROCm 10.0 software stack. The demo marks a notable step for the open RISC-V instruction set architecture in data center AI computing.

The companies said the collaboration combines open-source software, AMD's ROCm ecosystem, and the strengths of RISC-V's open standard architecture to enable next-generation data center AI workloads. They will continue evaluating ROCm optimizations for servers built on RISC-V, aiming to improve processing performance and support larger models and more accelerated use cases.

Matt Langman, SiFive's senior vice president of data center, said bringing AMD's open-source ROCm ecosystem to the BigSky data center development platform helps hyperscalers and developers run advanced AI workloads on RISC-V. Ramine Roane, AMD's vice president of AI software product management, noted that ROCm's open foundation allows developers to bring accelerated AI workloads to a wide range of systems, calling the SiFive demo a first step toward exploring ROCm-based AI acceleration on RISC-V host platforms.

Explore topicsAMDCompute PowerGemmaModelsSiFive

Source

View all