2026-06-09 –, Poster Island C
To achieve both high performance and productivity, modern software stacks rely on linear algebra libraries that encapsulate decades of optimization efforts. These libraries derive from empirical studies and analytical models developed primarily for high-end general-purpose systems, particularly the vector/SIMD extensions of x86-64 processors (e.g., SSE, AVX, and AVX-512). As a new and open-standard architecture, RISC-V software programming models target systems with unprecedent hardware diversity since pivotal extensions, such as the RISC-V V, may be realized on processors for domains ranging from edge to supercomputing. In this work, we advocate for flexible code generation tools to foster vendor-agnostic performance on high-performance software stacks, highlighting the most impactful software-level performance optimizations on early RISC-V vector systems and our insights on how to handle them on the context of linear algebra library development.
Alexandre de Limas Santana has been working on high-performance code generation at the Barcelona Supercomputing Center for the past five years. During this time, he participated in the European Pilot project as a research engineer, where he focused on accelerating dense linear algebra operations for deep neural networks on long-vector processors. Alexandre’s research interests now also include performance portability, motivated by the hardware diversity of emerging RISC-V vector architectures.