vlang/vsl

V library to develop Artificial Intelligence and High-Performance Scientific Computations

What it solves

VSL provides a high-performance foundation for AI and scientific computing in the V language. It fills the gap for low-level mathematical primitives—such as linear algebra, numerical methods, and GPU acceleration—that are necessary to build complex machine learning models and scientific simulations.

How it works

VSL implements a tiered performance architecture. It offers pure-V implementations of BLAS/LAPACK routines for portability and ease of installation. For maximum performance, users can opt-in to optimized backends including OpenBLAS, LAPACK, CUDA (cuBLAS/cuDNN), Vulkan Compute, and OpenCL/VCL. It uses a unified vsl.compute dispatch API to allow downstream libraries (like VTL for neural networks) to access these hardware-accelerated primitives without needing to rewrite their logic for every backend.

Who it’s for

It is designed for developers using the V language who need to perform heavy numerical computation, build machine learning algorithms from scratch, or create scientific applications requiring GPU acceleration.

Highlights

  • Comprehensive Math Suite: Includes linear algebra (eigenvalue decomposition, solvers), numerical methods (differentiation, integration), and statistical analysis.
  • GPU Acceleration: Native support for CUDA, Vulkan, and OpenCL to handle computationally intensive operations.
  • ML Primitives: Provides the underlying compute kernels (GEMM, Conv2D, softmax, LayerNorm) used by neural network libraries.
  • Flexible Backends: Allows switching between pure-V, C-based, and GPU-accelerated implementations via compilation flags.
  • Scientific Data Support: Integration with HDF5 for handling large scientific datasets.

Related

  • Project
    NVIDIA/cuvs

    A GPU-accelerated library for vector search and clustering, providing high-performance approximate nearest neighbors implementations for semantic search and data mining.

  • Project
    NVIDIA/cuml

    A GPU-accelerated machine learning library that provides scikit-learn-style APIs and a transparent acceleration module to speed up ML workloads on NVIDIA GPUs.

  • Project
    uxlfoundation/oneDAL

    A C++ and DPC++ library that provides hardware-accelerated machine learning routines for tabular data across CPUs, GPUs, and distributed systems.

  • Project
    SciML/DifferentialEquations.jl

    A high-performance numerical solver suite for a wide variety of differential equations, supporting scientific machine learning and GPU acceleration.

  • Project
    ROCm/rocBLAS

    rocBLAS is a Basic Linear Algebra Subprograms library implemented in HIP and optimized for AMD GPUs to provide high-performance linear algebra operations.