Nvidia cuda toolkit windows — Ваш верный помощник с OS Windows

Resources

CUDA Documentation/Release Notes
MacOS Tools
Training
Sample Code
Forums

Archive of Previous CUDA Releases
FAQ
Open Source Packages
Submit a Bug
Tarball and Zip Archive Deliverables

The NVIDIA® CUDA® Toolkit provides a development environment for creating high-performance, GPU-accelerated applications. With it, you can develop, optimize, and deploy your applications on GPU-accelerated embedded systems, desktop workstations, enterprise data centers, cloud-based platforms, and supercomputers. The toolkit includes GPU-accelerated libraries, debugging and optimization tools, a C/C++ compiler, and a runtime library.

Download Now

The Features of CUDA 12

Built-In Capabilities for Easy Scaling

Using built-in capabilities for distributing computations across multi-GPU configurations, you can develop applications that scale from single-GPU workstations to cloud installations with thousands of GPUs.

Learn More

New Release, New Benefits

CUDA 12 introduces support for the NVIDIA Hopper™ and Ada Lovelace architectures, Arm® server processors, lazy module and kernel loading, revamped dynamic parallelism APIs, enhancements to the CUDA graphs API, performance-optimized libraries, and new developer tool capabilities.

Learn More

Support for Hopper

Support for the Hopper architecture includes next-generation Tensor Cores and Transformer Engine, the high-speed NVIDIA NVLink® Switch, mixed-precision modes, second-generation Multi-Instance GPU (MIG), advanced memory management, and standard C++/Fortran/Python parallel language constructs.

Learn More

Tutorials

CUDA Developer Tools is a series of tutorial videos designed to get you started using NVIDIA Nsight™ tools for CUDA development. It explores key features for CUDA profiling, debugging, and optimizing.

GTC Digital Webinars

Dive deeper into the latest CUDA features.

Inside the NVIDIA Hopper Architecture

Explore what’s new with the NVIDIA Hopper architecture and its implementation in the NVIDIA H100 Tensor Core GPU.

Watch Now

CUDA—New Features and Beyond

Learn what’s new in the CUDA Toolkit, including the latest and greatest features in the CUDA language, compiler, libraries, and tools—and get a sneak peek at what’s coming up over the next year.

Watch Now

CUDA on NVIDIA Hopper GPU Architecture

Learn how to leverage the NVIDIA Hopper architecture’s capabilities to take your algorithms to the next level of performance.

Watch Now

See All

Customer Stories

See how developers, scientists, and researchers are using CUDA today.

Using HPC to Explore the Universe

Wes Armour, director at the Oxford e-Research Centre, discusses the role of GPUs in processing large amounts of astronomical data collected by the Square Kilometre Array and how CUDA is the best-suited option for their signal processing software.

Watch Video

Opening a New Era of Drug Discovery With Amber

David Cerutti and Taisung Lee from Rutgers University share how Amber, harnessing CUDA, is advancing multiple scientific domains and opening a new era of drug discovery and design.

Watch Video

Visualizing and Simulating Atomic Structures

John Stone, senior research programmer at the Beckman Institute at the University of Illinois, Urbana-Champaign, discusses how CUDA and GPUs are used to process large datasets to visualize and simulate high-resolution atomic structures.

Watch Video

CUDA Ecosystem

Explore the top compute and graphics packages with built-in CUDA integration.

Featured Blogs

Latest News

Free Tools and Trainings for Developers

Get exclusive access to hundreds of SDKs, technical trainings, and opportunities to connect with millions of like-minded developers, researchers, and students.

Learn more

Resources

CUDA Documentation and Release Notes

Documentation library containing in-depth technical information on the CUDA Toolkit.

CUDA 12 Features Revealed

A technical blog on the CUDA Toolkit 12.0’s features and capabilities.

CUDA Toolkit in the NGC Catalog

CUDA containers are available to download from NGC™—along with other NVIDIA GPU-accelerated SDKs and AI models—to help accelerate your applications.

Learn more

All CUDA Technical Blogs

An archive of CUDA technical blogs covering key features and capabilities, written by engineers for engineers.

Learn more

CUDA-X™ Libraries

A suite of AI, data science, and math libraries developed to help developers accelerate their applications.

Learn more

Training

Self-paced or instructor-led CUDA training courses for developers through the NVIDIA Deep Learning Institute (DLI).

Learn more

Nsight Developer Tools

NVIDIA Nsight Compute and Nsight System suite of tools designed to help developers optimize and increase performance of their applications.

Learn more

Sample CUDA Code

GitHub repository of sample CUDA code to help developers learn and ramp up development of their GPU-accelerated applications.

NVIDIA Developer Forums

An information exchange to help developers get answers to their technical questions directly from NVIDIA engineers.

Bug Submission

NVIDIA Engineering’s own bug tracking tool and database where developers can submit technical bugs.

Learn more

Источник

The NVIDIA CUDA Toolkit is an essential software platform for anyone looking to unlock the immense parallel processing power of NVIDIA GPUs.

Whether you are a researcher leveraging GPU-acceleration for cutting-edge deep learning or a developer harnessing GPU computing for simulations, 3D rendering, and other computational workloads, installing the CUDA Toolkit is the first step to supercharge your work.

In this comprehensive guide, we will cover everything you need to know to properly install the latest version of the NVIDIA CUDA Toolkit on Linux, Windows and macOS systems.

What is NVIDIA CUDA Toolkit?

The NVIDIA CUDA Toolkit provides a development environment for creating high performance GPU-accelerated applications. It allows developers to harness the parallel processing capabilities of NVIDIA GPUs for significant performance improvements compared to CPU-only workflows.

Here are some of the key components included in the CUDA Toolkit:

CUDA Driver API and Runtime: This enables direct access to the GPU’s virtual instruction set and parallel computational elements.
Compilers: NVCC compiler for CUDA C/C++ programming. OpenACC, OpenMP, and MPI support.
Math Libraries: BLAS, FFT, RNG, and other GPU-accelerated math libraries.
Development Tools: NVIDIA Nsight IDE, debugger, profiler and more.
Code Samples and Documentation: Everything you need to start CUDA development.

By providing essential GPU acceleration enablers like the CUDA parallel computing platform and programming model, the CUDA Toolkit empowers developers to solve complex computational challenges faster and more efficiently.

Key Benefits of the CUDA Toolkit

Achieve massive performance improvements with parallel processing on GPUs.
Write CUDA C/C++ code for GPU acceleration without specialized skills.
Port existing C/C++ code to run on GPUs.
Analyze and optimize CUDA applications for maximal efficiency.
Develop, debug and profile seamlessly in familiar environments.

Who Should Install the CUDA Toolkit?

The CUDA Toolkit is designed for software developers, researchers and data scientists who want to leverage NVIDIA GPUs for high performance computing and deep learning applications.

Whether you are accelerating a computational fluid dynamics simulation, training neural networks, running molecular dynamics simulations or deploying any other GPU-accelerated workload, installing the CUDA Toolkit is the first step to harness advanced parallel computing capabilities.

Now that you understand the immense value that the CUDA Toolkit delivers, let’s get into the specific steps you need to follow to properly install it on your system.

How to Install NVIDIA CUDA Toolkit?

The CUDA Toolkit is supported on most modern Linux distributions, Windows 7 or later and macOS 10.13 or later. I will cover the detailed instructions for installation on these operating systems.

Linux Installation

Most Linux distributions include CUDA in their package manager repositories. This makes installing the CUDA Toolkit very straightforward.

Here are the steps to install the latest version of the CUDA Toolkit on Linux:

Step 1: Verify System Requirements

A desktop or workstation with NVIDIA GPU with CUDA compute capability 3.0 or higher.
64-bit Linux distribution with a glibc version later than 2.17. Consult the CUDA Installation Guide for specific version requirements.
gcc compiler and toolchain.
Up-to-date NVIDIA graphics drivers.

Step 2: Download the CUDA Toolkit

Go to the CUDA Toolkit download page: https://developer.nvidia.com/cuda-downloads
Choose the right package for your Linux distribution. For example, Ubuntu 18.04 would require:
cuda-repo-ubuntu1804-10-2-local-10.2.89-440.33.01_1.0-1_amd64.deb
Download the installer to your machine.

Step 3: Install the CUDA Toolkit

Open a terminal and navigate to the download directory.
Install the downloaded package with sudo dpkg -i [package name].
Follow the on-screen prompts. Accept EULA and install the CUDA Toolkit components.
The installation process will automatically attempt to install the NVIDIA graphics driver if a CUDA compatible version is not already present.

Step 4: Verify the Installation

To verify that CUDA is installed and working correctly, compile and run a CUDA sample program:

cd /usr/local/cuda/samples/1_Utilities/deviceQuery
make
./deviceQuery

This runs a small CUDA program to verify CUDA capabilites. If installed correctly, it will show detected NVIDIA GPUs and capabilities.

The CUDA Toolkit is now installed and ready to use! You can start writing your own CUDA programs or porting existing computational workloads to run on NVIDIA GPUs.

Windows Installation

On Windows platforms, the most straightforward method to install the CUDA Toolkit is by using the standalone Windows installer from NVIDIA.

Follow these steps for smooth installation on Windows:

Step 1: Verify System Requirements

A desktop or notebook PC with NVIDIA GPU with CUDA compute capability 3.0 or higher.
64-bit version of Windows 7 or later. Windows 10/11 recommended.
Visual Studio IDE installed. Visual Studio 2019 recommended.
Latest NVIDIA graphics driver compatible with CUDA.

Step 2: Download the CUDA Toolkit

Get the Windows CUDA Toolkit installer from:
https://developer.nvidia.com/cuda-downloads
Choose the exe network installer for Windows x86_64.
Download the installer to your machine.

Step 3: Install the CUDA Toolkit

Double click the downloaded exe installer file.
Click through the NVIDIA license agreement. Select accept and continue.
Select the components to install:
CUDA Toolkit
CUDA Samples
Visual Studio Integration (optional but recommended)
NVIDIA Display Driver (if compatible version not already installed)
Click Install to begin the installation process.

Step 4: Verify the Installation

Launch Visual Studio and create a new CUDA C/C++ project.
Try compiling and running one of the CUDA samples.
For example, the deviceQuery sample prints information about the CUDA-enabled GPUs on the system.

With the CUDA Toolkit properly installed, you can commence CUDA application development on the Windows platform.

macOS Installation

On macOS, the CUDA Toolkit is provided as a DMG installer image containing the toolkit, samples and necessary drivers.

Follow these instructions to install the CUDA Toolkit on macOS:

Step 1: Verify System Requirements

Mac with NVIDIA GPU based on:
Maxwell or newer GPU architecture.
CUDA compute capability 5.0 and higher.
macOS High Sierra 10.13 or later.
Latest matching NVIDIA macOS graphics driver.

Step 2: Download the CUDA Toolkit

Get the macOS CUDA Toolkit DMG installer from:
https://developer.nvidia.com/cuda-downloads
Choose the macOS installer DMG package.
Download the installer image.

Step 3: Install the CUDA Toolkit

Double click the DMG installer package to mount it.
Double click the mounted volume icon to open it.
Double click the CUDA-Install-macOS.pkg file to launch the installer.
Click continue and accept the license agreement.
Follow the prompts to install the CUDA Toolkit, Samples and Driver components.

Step 4: Verify the Installation

Open a terminal and compile and run a CUDA sample like deviceQuery:

cd /Developer/NVIDIA/CUDA-X.Y/samples/1_Utilities/deviceQuery
make
./deviceQuery

This will confirm CUDA is correctly set up if the system’s NVIDIA GPU is detected.

With the CUDA Toolkit installed, your macOS system is now ready for serious GPU computing!

Additional Notes on CUDA Toolkit Installation

Here are some additional pointers to ensure smooth installation and operation of the CUDA Toolkit:

When installing CUDA on a workstation or server running Linux, it is generally recommended to use the *.run package installer instead of distro-specific packages.
On Linux/Windows, install the cuDNN libraries after CUDA for GPU acceleration of deep neural networks.
Setup the PATH and LD_LIBRARY_PATH environment variables to point to the CUDA Toolkit install location.
For developing CUDA applications, installing a compatible version of the NVIDIA Nsight IDE or Visual Studio IDE is highly recommended.
When deploying CUDA applications on cloud platforms like AWS EC2 or Azure NV-series VMs, follow NVIDIA’s guides to install CUDA.
Refer to NVIDIA’s documentation for troubleshooting guidance, advanced installation options, and other details specific to your system configuration.

Learn More and Get Started with CUDA Programming

With the CUDA Toolkit properly installed, an exciting world of GPU-accelerated computing is now open to you.

Some helpful next steps:

Try out the CUDA Sample programs included in the Toolkit installation.
Walk through the CUDA Programming Guide for a comprehensive tutorial.
Refer to the official CUDA Documentation for API references and expert guides.
Join the CUDA Developer Forums to engage with the CUDA community.
Stay updated on the NVIDIA Developer Blog covering cutting-edge GPU computing applications.

I hope this detailed guide helped demystify the process for installing the NVIDIA CUDA Toolkit on your system. The remarkable acceleration and performance benefits unlocked by CUDA are now at your fingertips. Happy coding!

FAQ’s

Where is Nvidia CUDA toolkit installed?

The Nvidia CUDA toolkit is typically installed in the /usr/local/cuda directory on Linux, under C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA on Windows, and in /Developer/NVIDIA/CUDA-X.Y on macOS.

Does Nvidia Cuda Toolkit install drivers?

Yes, the Nvidia CUDA toolkit installer can optionally install Nvidia graphics drivers if a compatible version is not already present on the system. It is recommended to install the latest drivers matched with your CUDA version.

How to install CUDA toolkit without Sudo?

To install the CUDA toolkit without Sudo access on Linux, download the runfile installer and execute it with the –tmpdir option pointing to a writable local directory. This will install CUDA toolkit components in your user folder.

Is CUDA and CUDA toolkit the same?

CUDA refers to Nvidia’s parallel computing platform and API. The CUDA toolkit is the software development package from Nvidia that provides libraries, compiler, tools and samples to build CUDA applications.

How do I enable CUDA?

To enable CUDA, install a compatible Nvidia graphics driver, install the CUDA toolkit, configure the PATH and LD_LIBRARY_PATH to include the CUDA install directories, and verify by running CUDA sample programs.

Is CUDA a CPU or GPU?

CUDA is Nvidia’s API and platform to utilize the parallel processing capabilities of Nvidia GPUs. It allows compute intensive tasks to be offloaded from the CPU to the GPU.

Do all GPUs have CUDA?

No, only Nvidia GPUs designed for general purpose GPU computing support the CUDA platform. AMD and Intel GPUs do not support CUDA.

Can CUDA run on AMD graphics?

No, CUDA only runs on Nvidia GPUs. For AMD GPUs, OpenCL is the alternative to CUDA for GPGPU computing

Источник

Download CUDA® installer

You can download the CUDA® toolkit for your operating system from the NVIDIA® Developers Portal. We recommend using the local installer type, as it can be faster. The installer is quite large, with a size of about 3 GB. However, each LeaderGPU server has a very fast internet connection, so it doesn’t take a lot of time:

Open downloads in Chrome by pressing the Ctrl + J keyboard shortcut and double-clicking on the downloaded CUDA® toolkit installer:

Run CUDA® installer

Most NVIDIA® packages are self-extracted archives with installers inside. You can select a specific folder or leave the default path, then click OK:

Wait a minute while the archive is being extracted:

The installer will begin. In the first stage, the installer checks your hardware compatibility with the system requirements:

To proceed, you must agree with EULA by clicking AGREE AND CONTINUE button:

Select Express to install all available components and click NEXT:

Some tools integrate with the Visual Studio IDE. If it isn’t installed on the server, the installer will issue a warning. Check the box and click NEXT:

Installation may take a few minutes, which is enough time to pour yourself some coffee:

After installation, you can read the summary and click NEXT:

Finally, you can uncheck boxes (creating shortcut and launch utility) or leave them as they are. Exit the installer by clicking on the CLOSE button:

Support the Project ⭐

If you find this project useful, please give it a star! Your support is appreciated and helps keep the project growing. 🌟

🚀 NVIDIA CUDA Installation Guide

This guide walks you through installing NVIDIA CUDA Toolkit 11.8, cuDNN, and TensorRT on Windows, including setting up Python packages like Cupy and TensorRT. It ensures proper system configuration for CUDA development, with steps for setting environment variables and verifying installation via cmd.exe

1. Download the NVIDIA CUDA Toolkit 11.8

First, download the CUDA Toolkit 11.8 from the official NVIDIA website:

👉 Nvidia CUDA Toolkit 11.8 — DOWNLOAD HERE

2. Install the CUDA Toolkit

After downloading, open the installer (.exe) and follow the instructions provided by the installer.
Make sure to select the following components during installation:
- CUDA Toolkit
- CUDA Samples
- CUDA Documentation (optional)

3. Verify the Installation

After the installation completes, open the cmd.exe terminal and run the following command to ensure that CUDA has been installed correctly:

This will display the installed CUDA version.

4. Install Cupy

Run the following command in your terminal to install Cupy:

5. CUDNN Installation 🧩

Download cuDNN (CUDA Deep Neural Network library) from the NVIDIA website:

👉 Download CUDNN. (Requires an NVIDIA account – it’s free).

6. Unzip and Relocate 📁➡️

Open the .zip cuDNN file and move all the folders/files to the location where the CUDA Toolkit is installed on your machine, typically:

C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8

7. Get TensorRT 8.6 GA 🔽

Download TensorRT 8.6 GA.

8. Unzip and Relocate 📁➡️

Open the .zip TensorRT file and move all the folders/files to the CUDA Toolkit folder, typically located at:

C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8

9. Python TensorRT Installation 🎡

Once all the files are copied, run the following command to install TensorRT for Python:

pip install "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\python\tensorrt-8.6.1-cp311-none-win_amd64.whl"

🚨 Note: If this step doesn’t work, double-check that the .whl file matches your Python version (e.g., cp311 is for Python 3.11). Just locate the correct .whl file in the python folder and replace the path accordingly.

10. Set Your Environment Variables 🌎

Add the following paths to your environment variables:

C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\lib
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\libnvvp
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\bin

Setting Up CUDA 11.8 with cuDNN on Windows

Once you have CUDA 11.8 installed and cuDNN properly configured, you need to set up your environment via cmd.exe to ensure that the system uses the correct version of CUDA (especially if multiple CUDA versions are installed).

Steps to Set Up CUDA 11.8 Using `cmd.exe`

1. Set the CUDA Path in `cmd.exe`

You need to add the CUDA 11.8 binaries to the environment variables in the current cmd.exe session.

Open cmd.exe and run the following commands:

set PATH=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\bin;%PATH%
set PATH=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\libnvvp;%PATH%
set PATH=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\extras\CUPTI\lib64;%PATH%

These commands add the CUDA 11.8 binary, lib, and CUPTI paths to your system’s current session. Adjust the paths as necessary depending on your installation directory.

Verify the CUDA Version
After setting the paths, you can verify that your system is using CUDA 11.8 by running:

This should display the details of CUDA 11.8. If it shows a different version, check the paths and ensure the proper version is set.

Set the Environment Variables for a Persistent Session
If you want to ensure CUDA 11.8 is used every time you open cmd.exe, you can add these paths to your system environment variables permanently:
Open Control Panel -> System -> Advanced System Settings.
Click on Environment Variables.
Under System variables, select Path and click Edit.
Add the following entries at the top of the list:

C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\bin
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\libnvvp
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\extras\CUPTI\lib64

This ensures that CUDA 11.8 is prioritized when running CUDA applications, even on systems with multiple CUDA versions.

Set CUDA Environment Variables for cuDNN
If you’re using cuDNN, ensure the cudnn64_8.dll is also in your system path:

set PATH=C:\tools\cuda\bin;%PATH%

This should properly set up CUDA 11.8 to be used for your projects via cmd.exe.

Environmental Variable Setup

import torch

print(torch.cuda.is_available())  # This will return True if CUDA is available
print(torch.version.cuda)  # This will print the CUDA version being used
print(torch.cuda.get_device_name(0))  # This will print the name of the GPU, e.g., 'NVIDIA GeForce RTX GPU Model'

run the get_device.py to see if you installed it correctly

Cuda Requirements

run the cuda-requirements.bat after you get done with installion of nvidia.

@echo off
:: Batch script to install Python packages for CUDA 11.8 environment

echo MAKE SURE TO HAVE THE WHL DOWNLOADED BEFORE YOU CONTINUE!!!
pause
echo Click the link to download the WHL: press ctrl then left click with mouse
echo https://github.com/cupy/cupy/releases/download/v13.4.1/cupy_cuda11x-13.4.1-cp311-cp311-win_amd64.whl
pause

echo Installing CuPy from WHL...
pip install https://github.com/cupy/cupy/releases/download/v13.4.1/cupy_cuda11x-13.4.1-cp311-cp311-win_amd64.whl
echo Press enter to continue with the rest of the dependency installs
pause

echo Installing ONNX Runtime with GPU support...
pip install onnxruntime-gpu==1.19.2
echo Press enter to continue with the rest of the dependency installs
pause

echo Installing NVIDIA PyIndex...
pip install nvidia-pyindex
echo Press enter to continue with the rest of the dependency installs
pause

echo Installing cuDNN for CUDA 11.8...
pip install nvidia-cudnn-cu11==8.6.0.163
echo Press enter to continue with the rest of the dependency installs
pause

echo Installing TensorRT for CUDA 11.8...
pip install nvidia-tensorrt==8.6.1
echo Press enter to continue with the rest of the dependency installs
pause

echo Installing NumPy...
pip install numpy
echo Press enter to continue with the rest of the dependency installs
pause

echo Installing cupy-cuda11x...
pip install cupy-cuda11x
echo Press enter to continue with the rest of the dependency installs
pause

echo All packages installed successfully!
pause

Источник

Resources

The Features of CUDA 12

Built-In Capabilities for Easy Scaling

New Release, New Benefits

Support for Hopper

Tutorials

GTC Digital Webinars

Inside the NVIDIA Hopper Architecture

CUDA—New Features and Beyond

CUDA on NVIDIA Hopper GPU Architecture

Customer Stories

Using HPC to Explore the Universe

Opening a New Era of Drug Discovery With Amber

Visualizing and Simulating Atomic Structures

CUDA Ecosystem

Featured Blogs

Latest News

Free Tools and Trainings for Developers

Resources

CUDA Documentation and Release Notes

CUDA 12 Features Revealed

CUDA Toolkit in the NGC Catalog

All CUDA Technical Blogs

CUDA-X™ Libraries

Training

Nsight Developer Tools

Sample CUDA Code

NVIDIA Developer Forums

Bug Submission

What is NVIDIA CUDA Toolkit?

Who Should Install the CUDA Toolkit?

How to Install NVIDIA CUDA Toolkit?

Linux Installation

Windows Installation

macOS Installation

Additional Notes on CUDA Toolkit Installation

Learn More and Get Started with CUDA Programming

FAQ’s

Download CUDA® installer

Run CUDA® installer

Support the Project ⭐

🚀 NVIDIA CUDA Installation Guide

1. Download the NVIDIA CUDA Toolkit 11.8

2. Install the CUDA Toolkit

3. Verify the Installation

4. Install Cupy

5. CUDNN Installation 🧩

6. Unzip and Relocate 📁➡️

7. Get TensorRT 8.6 GA 🔽

8. Unzip and Relocate 📁➡️

9. Python TensorRT Installation 🎡

10. Set Your Environment Variables 🌎

Setting Up CUDA 11.8 with cuDNN on Windows

Steps to Set Up CUDA 11.8 Using cmd.exe

1. Set the CUDA Path in cmd.exe

Environmental Variable Setup

Cuda Requirements

Steps to Set Up CUDA 11.8 Using `cmd.exe`

1. Set the CUDA Path in `cmd.exe`