SiFive has announced its 2nd Generation Intelligence series, introducing five RISC-V-based intellectual property (IP) products engineered to accelerate artificial intelligence (AI) workloads across various platforms. The release includes two new IP cores, the X160 Gen 2 and X180 Gen 2, alongside upgraded versions of the X280 Gen 2, X390 Gen 2, and XM Gen 2. SiFive claims these products offer improved scalar, vector, and—specifically in the XM core—matrix processing capabilities tailored for modern AI applications.

The X160 Gen 2 and X180 Gen 2 IP cores are designed for far edge compute and Internet of Things (IoT) use cases, targeting scenarios with strict power and area constraints. According to SiFive, these cores are suitable for automotive, industrial automation, autonomous robotics, and smart IoT environments that require advanced AI functionality in compact, efficient silicon.

SiFive states that all Gen 2 IPs benefit from enhanced vector processing architectures. The company says vector engines enable AI model acceleration by processing multiple data elements in parallel, which can reduce instruction overhead and overall power consumption compared to scalar-only central processing unit (CPU) solutions. The new XM Gen 2 incorporates a matrix engine to further accelerate computation-heavy AI tasks.

For data center and platform-level AI workloads, SiFive claims its X-series IP can serve as an Accelerator Control Unit, interfacing with customers’ accelerator engines through specialized co-processor interfaces (SSCI and VCIX). This allows for architectural flexibility and platform-level data processing enhancements.

According to SiFive, all five Intelligence Gen 2 IP products are now available for licensing, with the first chips based on this IP expected to reach production in the second quarter of 2026.

Source: SiFive