VSL Documentation

June 1, 2026 · View on GitHub

Welcome to VSL, the scientific computing foundation for V and the low-level compute layer used by VTL for ML workloads.

Use this page as the stable navigation hub for tutorials, examples, GPU backend docs, benchmarks, and release status.

Start Here

Goal	Read
Install VSL and run the first example	Installation and Setup
Learn the core scientific APIs	Basic Linear Algebra, BLAS Basics
Build ML pipelines with VSL algorithms	Data Preparation, Clustering, Regression, Classification
Train tensor/autograd neural networks	VTL
Use GPU backends	OpenCL, CUDA, Vulkan
Compare performance	Benchmarks, vs NumPy
Track the ML release	ML Roadmap

Start here if you're new to VSL or scientific computing:

Build on the basics with more advanced topics:

Master advanced techniques and integrations:

Essential tutorials for beginners:

Matrix operations, BLAS, and LAPACK:

Creating plots and charts:

3D rotations and orientations:

ML algorithms and data processing:

Numerical methods and signal processing:

High-performance and specialized topics:

Backend	Build flag	Current role	Docs
Pure V	none	Portable scientific computing	README
C BLAS/LAPACK	`-d vsl_blas_cblas`, `-d vsl_lapack_lapacke`	Optimized CPU linear algebra	Linear algebra
OpenCL/VCL	module-specific	Cross-vendor GPU kernels and examples	OpenCL GPU
CUDA	`-d cuda`	cuBLAS/cuDNN kernels for NVIDIA GPUs	CUDA README
Vulkan	`-d vulkan`	GEMM, Conv2D im2col, elementwise ops, fused Adam	Vulkan README

VSL owns compute kernels and backend dispatch. For tensors, autograd, datasets, layers, losses, optimizers, and training examples, use VTL.

Create a plot:

import vsl.plot

mut plt := plot.Plot.new()
plt.scatter(x: [1, 2, 3], y: [4, 5, 6])
plt.show()!

Matrix multiplication:

import vsl.blas
// Use BLAS dgemm for matrix multiplication

Quaternion rotation:

import vsl.quaternion
import math

q := quaternion.from_axis_anglef3(math.pi / 2, 1, 0, 0)

All tutorials include links to working examples in the examples/ directory. Each example includes:

For a curated catalog, see examples/README.md.

Found an error or want to improve a tutorial? See our Contributing Guide.