The SiFive Performance P870 and SiFive Intelligence X390 provide a new level of low power, compute density, and vector compute capability, and when combined can provide a performance boost for increasingly data intensive compute.
Together, the new products create a powerful mix of scalar and vector computing to address the needs of dataflow and computation intensive AI applications across consumer, automotive, and infrastructure markets.
“SiFive is leading the industry into a new era of high-performance RISC-V innovation and closing the gap with other instruction set architectures with our unparalleled portfolio, while recent silicon tape-outs are demonstrating the tremendous benefits of SiFive RISC-V solutions,” said Patrick Little, SiFive Chairman, President and CEO. “As the Arm IPO showed, there is a fast-growing demand for semiconductors across many sectors, particularly processors for consumer and infrastructure markets. The flexibility of SiFive’s RISC-V solutions allows companies to address the unique computing requirements of these segments and capitalise on the momentum around generative AI, where we have seen double-digit design wins, and for other cutting-edge applications.”
Intended for high performance consumer applications, or when used in conjunction with a vector processor in the datacentre, the P870 core has been able to set an impressive new RISC-V performance bar across instruction set architecture availability, throughput, parallelism, and memory bandwidth. Bringing a 50% peak single thread performance upgrade (specINT2k6) over the previous generation SiFive Performance processors, the P870 is a six-wide out-of-order core, that meets RVA 23 and offers a shared cluster cache enabling up to a 32-core cluster.
High execution throughput comes with more instruction sets per cycle, more ALU, and more branch units. The P870 is fully compatible with Google’s platform requirements for Android on RISC-V. The P870 also offers additional proven SiFive features: · x 128b VLEN RVV · Vector crypto and hypervisor extensions · IOMMU and AIA · Non-inclusive L3 cache · Proven RISC-V WorldGuard security
Building on the highly popular SiFive Intelligence X280’s success in coupling AI/ML applications with hardware accelerators in mobile, infrastructure, and automotive applications, the new X390 brings a 4x improvement to vector computation with its single core configuration, doubled vector length, and dual vector ALUs. This allows quadruple the amount of sustained data bandwidth.
With SiFive Vector Coprocessor Interface eXtension (VCIX) companies can easily add their own vector instructions and/or acceleration hardware, bringing unprecedented flexibility and allowing users to greatly increase performance with custom instructions. Features include: · 1024-bit VLEN, 512-bit DLEN · Single / Dual Vector ALU · VCIX (2048-bit out, 1024-bit in)
Bringing the P870 high-performance general compute SoC together with a high performance NPU cluster, consisting of the X390 and customer AI hardware engines, offers product designers a more flexible, low power, and programmable solution with superior compute density for complex workloads.
