Scalable Matrix Extension: Expanding the Arm Intrinsics Search Engine

Arm is pleased to announce that the Arm Intrinsics Search Engine has been updated to include the Scalable Matrix Extension (SME) intrinsics, including both SME and SME2 intrinsics.

The Arm Intrinsics Search Engine is designed to save you time and reduce complexity when working with Arm intrinsics. By providing a unified resource with clear, searchable information, it enables quick discovery and understanding of how intrinsics map to underlying instructions.

SME extends the Arm architecture with advanced matrix-computation capabilities. These capabilities accelerate high-performance computing and machine learning workloads.
The search engine now provides information about the complete set of Arm intrinsics:

• SME
• SVE and SVE2
• Neon
• Helium

You can search the SME intrinsics by function name, Arm instruction, or description. The results show detailed reference information such as function prototypes, supported architectures, and links to related instructions.

Explore the updated intrinsics search engine today:

Arm Intrinsics Search Engine

What are intrinsics?

Intrinsics are special compiler functions. They give developers direct access to specific instructions without writing assembly code. They bridge between high-level languages like C or C++ and the low-level capabilities of the processor. This lets developers write portable, optimized code that harnesses advanced hardware features.

Intrinsics are important for performance-critical applications. They allow fine-grained control over vector, matrix, and other specialized instructions that can significantly accelerate computation.

What is SME?

Arm Scalable Matrix Extension (SME) is an architecture extension that provides enhanced support for matrix operations. SME builds on the Scalable Vector Extensions (SVE and SVE2). It adds new capabilities to process matrices efficiently. Key features of SME include:

• Outer product between two vectors
• Matrix tile storage
• Load, store, insert, and extract of tile vectors, including on-the-fly transposition
• Streaming SVE mode

Read the Arm’s New Lumex CSS Platform news article to discover how SME2-enabled Arm CPUs are delivering up to 5x faster AI performance.

Get started

You can explore the updated intrinsics search engine today at:

• Arm Intrinsics Search Engine

For more information about SME, see:

• Developer Launchpad: Accelerate Mobile AI on Arm With SME2
• Arm Scalable Matrix Extension (SME) Introduction
• Arm Architecture Reference Manual for A-profile architecture
• SME Programmer's Guide

For information about using intrinsics, see:

• Arm SIMD Extensions Best Practice
• Learn the architecture - Optimizing C code with Neon intrinsics
• Neon Intrinsics - Getting Started on Android
• Optimizing Image Processing with Neon Intrinsics