Arm Compute Library (ACL) is a key component of Arm Kleidi, which brings together the latest developer enablement technologies and critical developer resources to accelerate AI development and enhance performance across Arm-based platforms. ACL provides a comprehensive set of low-level machine learning functions optimized for Arm Cortex-A CPU, Arm Neoverse, and Arm Mali GPU architectures, to facilitate superior ML performance. The library is open-source software available under a permissive MIT license.
The Arm Compute Library provides superior performance to other open-source alternatives and immediate support for new Arm technologies and architecture features, including SVE2 and SME2.
Key Features:
- Over 100 machine learning functions for CPU and GPU
- Multiple convolution algorithms (GEMM, Winograd, FFT and Direct)
- Support for multiple data types: FP32, FP16, int8, uint8, BFloat16
- Micro-architecture optimization for key ML primitives
- Highly configurable build options enabling lightweight binaries
- Advanced optimization techniques such as Kernel Fusion, fast math enablement and texture utilization
- Device-and workload-specific tuning using the Open CL tuner and GEMM-optimized heuristics
