SiFive’s BigSky server puts open-standard RISC-V hardware in the hands of datacentre developers, with workloads, RHEL 10 and CUDA already running on the platform.
SiFive has launched the BigSky SF-2U870, a rack-mount RISC-V development server designed to help hyperscalers, software developers and SoC vendors port, optimise and validate workloads for datacentre deployment.
Built around RISC-V, an open instruction-set architecture, the platform is intended to move the technology beyond experimentation towards production workloads. SiFive says lead customers are already running workloads on deployed BigSky systems.
The company claims BigSky is the world’s first enterprise-grade, rack-mount 2U RISC-V server, although that claim is difficult to independently verify. The system features 32 SiFive P870-D cores at 2 GHz, 256 GB DDR5 memory, two 7.68 TB NVMe SSDs and a 10/25 Gbps OCP-compliant network adapter.
SiFive Chairman and CEO Patrick Little said, “Ultimately, we are focused on enabling customers to meaningfully optimize the TCO of their custom SoC solutions with SiFive IP. RISC-V in the datacenter isn’t a distant aspiration any more, it is happening right now.”
SiFive is working with Canonical and Red Hat, with RHEL 10 running on BigSky. Red Hat Technology Strategist Brian Harrington said, “Getting RHEL 10 running on SiFive’s BigSky platform isn’t just a validation exercise, it’s proof that we’re moving from the ‘theoretical’ phase into production-grade reality.”
SiFive is also porting Nvidia CUDA to RISC-V and integrating NVLink Fusion into future designs. It says CUDA now runs on BigSky, potentially enabling RISC-V-based AI servers using Nvidia GPU clusters.
BigSky is available now in limited quantities.
