Skip to main content

Ctrl+K

cuBLASDx

cuBLASDx

Table of Contents

Documentation Home

Getting Started

Requirements
Installation
Using cuBLASDx GEMM
Using Pipelined GEMM
Using cuBLASDx TRSM
Achieving High Performance

API Reference

Operators
Traits
Input / Output Format
GEMM Execution Methods
TRSM Execution Methods
Pipeline Creation
Other Methods
Tensor Manipulation
Shared Memory Management

Python

cuBLASDx Python Bindings

Examples

Learning Path Through Examples
Examples

Agent BLAS Performance Guide

Overview
Porting An Existing Kernel
Finding Problem Shape
Fitting Into cuBLASDx Idioms
Finding Best Coarse Solution
Finetuning Your Program
API Card: Hopper Pipeline (wgmma)
API Card: Blackwell Pipeline (tcgen05)
API Card: supermma

References

Release Notes
GitHub Samples
License

Related Projects

libmathdx
NPPDx

MathDx

cuBLASDx
cuFFTDx
cuSolverDx
cuRANDDx
nvCOMPDx

Index

Privacy Policy | Your Privacy Choices | Terms of Service | Accessibility | Corporate Policies | Product Security | Contact

Copyright © 2022-2026, NVIDIA Corporation & Affiliates. All rights reserved.