Analyst(s): Dr. Bob Sutor, Steven Dickens
Publication Date: August 16, 2024
The News: SiFive announced the SiFive Performance P870-D, a high-performance RISC-V datacenter processor designed for demanding general and AI workloads, offering up to 256 cores, enhanced scalability, and advanced Reliability, Availability, and Serviceability (RAS) features. This processor aims to provide data center architects with a flexible, power-efficient solution for AI-driven applications, positioning SiFive as a competitive player in the data center market.
New SiFive Performance P870-D Complements Intelligence Chipset for AI
Analyst Take: RISC-V, an open-source instruction set architecture (ISA), has steadily gained traction in the global processor market, driven by its openness, flexibility, scalability, and the growing need for customizable solutions in various industries. Unlike proprietary ISAs such as those from Intel and AMD, RISC-V allows companies to develop processors tailored to specific workloads without being locked into a particular vendor’s ecosystem. This flexibility has made RISC-V an attractive option worldwide for companies looking to innovate rapidly, especially in areas such as artificial intelligence (AI), machine learning (ML), and edge computing.
Under CEO Patrick Little, SiFive has emerged as a key player in the RISC-V space, capitalizing on the momentum of the open-source movement. Since taking the helm, Little has focused on expanding SiFive’s ecosystem partnerships and enhancing the company’s product offerings to meet the demands of increasingly complex and data-intensive applications. He has positioned SiFive as a pioneer in the commercial RISC-V movement, helping to transform the architecture from a niche technology into a viable alternative for mainstream computing needs, particularly in AI-driven industries.
What Was Announced?
On August 14, 2024, SiFive announced its latest innovation: the SiFive Performance P870-D datacenter processor, a high-performance RISC-V processor designed to handle the most demanding AI workloads. This new processor is part of SiFive’s broader strategy to expand its footprint in the data center market, which established players such as Intel and AMD currently dominate.
The P870-D builds on the success of its predecessor, the P870, and introduces several enhancements that make it particularly well suited for AI-driven applications. Key among these enhancements is its support for up to 256 cores, a feature that significantly boosts compute density while maintaining power efficiency. This scalability allows data center architects to design systems that can handle a high volume of parallel tasks, making the P870-D ideal for workloads such as video streaming, storage, and web appliances.
One of the standout features of the P870-D is its support for the open AMBA CHI protocol, which offers customers greater flexibility in scaling the number of processor clusters. This, combined with support for industry-standard protocols such as Compute Express Link (CXL) and CHI chip-to-chip (C2C) communication, enables the creation of heterogeneous system-on-chip (SoC) and chiplet configurations. These configurations are increasingly important in the AI era, where integrating diverse processing elements into a single, coherent system is critical for optimizing performance and power consumption.
The P870-D’s sustainability and performance improvements work well with SiFive’s IntelligenceTM X280 chipset and its multi-core capable RISC-V processor with vector extensions. These extensions are beneficial for parallel processing of datasets and AI/ML computations at the edge. This is an important distinction for SiFive. Rather than focusing solely on the cloud and data centers, SiFive also enables AI for “AR, VR, sensor hubs, IVI systems, IP cameras, digital cameras, gaming devices,” per its website.
The P870-D includes advanced RAS features, making it suitable for functional safety applications in industries such as automotive, another edge use case. SiFive designed these RAS features to detect and correct errors proactively, thereby protecting data integrity and minimizing system downtime.
The processor supports a 57-bit virtual address space, particularly relevant for cloud computing environments requiring large memory spaces. Most major Linux distributions support the P870-D.
The RISC-V Story
RISC is short for “Reduced Instruction Set Computer.” Rather than have complicated low-level instructions, the RISC philosophy began in the late 1970s by observing that software compilers could combine simple instructions and allow better optimizations. Although you might have more instructions implementing operations, they can each be tuned for speed and sequenced in ways that make the overall process faster. One of the best-known implementations of RISC is the IBM Power architecture.
RISC-V is an open standard for a RISC ISA licensed under a BSD License by RISC-V International, based in Switzerland. The specification initially began as academic work at the University of California, Berkeley, but now has broad commercial development support. SiFive was one of the 23 founding members in 2015.
Looking Ahead
As SiFive continues to expand its presence in the data center market, the company faces significant challenges, particularly in competing with industry giants such as AMD and Intel. Both companies have long-standing relationships with data center operators and have developed comprehensive ecosystems supporting a wide range of applications. SiFive will need to differentiate itself through the performance and efficiency of its processors and by building a robust ecosystem of software, tools, and partnerships that can support the deployment and optimization of its RISC-V-based solutions.
One of the most significant challenges SiFive will face is gaining the trust of data center operators, who have traditionally relied on x86-based architectures. Convincing these operators to adopt a relatively new and open-source architecture will require SiFive to demonstrate clear performance, cost, and energy efficiency advantages. Additionally, SiFive will need to continue investing in R&D to ensure its processors can keep pace with the rapidly evolving demands of AI and other data-intensive applications. They must prove and publicize they can play nicely with NVIDIA, Groq, and Cerebras Systems chipsets.
The potential rewards for SiFive are substantial. The total addressable market (TAM) for data center processors is enormous and continues to grow as AI reshapes industries across the globe. As more companies invest in AI-driven applications, the demand for processors that can efficiently handle these workloads will only increase. RISC-V’s open-source foundation and customizability for specific tasks give it a strong chance of capturing a significant market share.
As sustainability becomes a more pressing concern for companies and governments alike, SiFive’s focus on power efficiency could give it a competitive edge. The P870-D’s ability to deliver high performance while minimizing power consumption aligns with the industry’s growing emphasis on reducing the environmental impact of data centers or shifting that energy savings to additional computation.
Key Takeaway
The SiFive Performance P870-D processor represents a significant step forward in the evolution of RISC-V architecture, bringing new levels of performance, scalability, and power efficiency to the data center market. By addressing the specific needs of AI workloads, SiFive is positioning itself as a formidable competitor in a market that more established players have traditionally dominated.
While SiFive faces formidable competition, its innovative approach to processor design and commitment to the RISC-V architecture position it well for future growth. As AI continues to disrupt the computing landscape, SiFive’s P870-D processor could play a crucial role in shaping the next generation of data center infrastructure.
For additional details, see the “SiFive Announces New High-performance RISC-V Datacenter Processor for Demanding AI Workloads” press release.
Disclosure: The Futurum Group is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author has no equity position in any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of The Futurum Group as a whole.
Other Insights from The Futurum Group:
SiFive Bumps Up Performance for RISC-V HCP, AI Systems
SiFive X280 RISC-V Processor Gets Scalability, Trust Updates
SiFive Raises Series F, Sells Core Assets, Reaches Multi-Billion Dollar Valuation
Author Information
Dr. Bob Sutor is an expert in quantum technologies with 40+ years of experience. He is the accomplished author of the quantum computing book Dancing with Qubits, Second Edition. Bob is dedicated to evolving quantum to help solve society's critical computational problems.
Steven engages with the world’s largest technology brands to explore new operating models and how they drive innovation and competitive edge.
