SiFive Partners with AMD to Optimize ROCm on RISC-V Data Center Processors
SiFive and AMD demonstrated AI model execution using AMD's ROCm 10.0 software stack on a RISC-V host system paired with an AMD Radeon AI PRO R9700 GPU. Running the Google Gemma model, the demonstration marks a key step in bringing AMD's open AI software stack to RISC-V data center servers. This partnership connects the open-standard RISC-V architecture with AMD's open ROCm software ecosystem, offering hyperscalers and developers an alternative to proprietary hardware and software stacks. It accelerates the adoption of RISC-V as a viable host CPU architecture for data center AI workloads alongside x86 and ARM. The demonstration took place at the AI Infra Summit using SiFive's BigSky enterprise server development platform. Both companies plan to continue optimizing ROCm on RISC-V processors to increase processing speed and support larger AI models in the future.
## BACKGROUND
RISC-V is an open-standard instruction set architecture (ISA) that enables custom chip designs for cloud and edge compute without licensing fees. AMD's ROCm (Radeon Open Compute) is an open-source software platform providing drivers, compilers, and APIs required to deploy high-performance computing and AI workloads on AMD GPUs.