Welcome to the Accelerated Ccomputing Developer Program home page, where you can engage with NVIDIA to help you do great things with GPU Computing.

You will find some exclusive downloads, including pre-production releases which are not be available to the public.

Read more about the latest CUDA Toolkit features on the CUDA Toolkit Page

Reporting An Issue

Registered developers can now submit bugs and issues directly to NVIDIA engineering teams using our new bug submission process. Bugs submitted previously will have already been processed but will be listed in the views below.

File attachments are currently not supported - should you want to send us a file, please wait until you get your NVIDIA Bug ID, and then sent the file(s) to CUDAIssues@nvidia.com referencing that ID number.

We are experiencing some issues during submission and editing bugs - please wait until this is resolved before submitting your new issues, sorry for any inconvenience.

Submit a New Bug

To check the status and comments about your bugs please use the "My Bugs" link from the left menu of your My Account page.

Submitted bugs can be edited to allow you to provide additional information, and also respond to questions or comments from our Engineering teams by following the view bugs link above.

Developer Program Exclusive Downloads

The downloads accessible via this portal are confidential and are provided exclusively to Members of the NVIDIA Developer Program. Please do not re-distribute or post the links to these files or the files themselves.

If you are not logged in or do not have correct membership - you will be prompted to register

Double Precision Boys Function Implementation

This is a double-precision implementation of the Boys function of orders 0 through 50 for molecular computations with Gaussian basis functions. It requires CUDA 6.0 or later and compute capability 2.0 or higher.

Download

Maxwell Compute Architecture: Compatibility Guide

 The Maxwell Compute Architure will be the foundation of many of NVIDIA's GPUs. This document discusses code compatibility.

Download

Maxwell Compute Architecture: Tuning Guide

 The Maxwell Compute Architure will be the foundation of many of NVIDIA's GPUs. This document discusses how to tune your code to optimize for this new architecture.

Download

AMGX Solver Library Trial

This is a high performance, state-of-the-art library and includes a flexible solver composition system that allows a user to easily construct complex nested solvers and preconditioners. Read more about its features.

Download

CUDA SIMD-within-a-word functions

An include file which contains a collection of inline functions for processing byte and half-word data packed into 32-bit words. The functions are hardware accelerated on Kepler platforms. Efficient emulation code is provided for earlier platform so the functions are portable across all compute capabilities. The functionality provided should be useful for image processing tasks and has many application areas.

Download

CUDA Double-Double Precision Arithmetic (Updated April 2013)

Source code implementing double-double precision arithmetic functions which take double-double precision operands using a simple C-style interface. Functions include:

  • Negation
  • Addition
  • Subtraction
  • Multiplication
  • Division
  • Square Root

Any developer who requires precision beyond double precision will find this download very useful.

Download

CUDA Accelerated Linpack

Download this code for GPU accelerated Linpack from your TESLA Cluster. For LINUX 64bit and Fermi Class GPU

Download

CUDA Batch Solver (Updated June 2013)

This code provides an efficient solver and matrix inversion for small matrices, using partial pivoting.

Download

CUDA Compiler SDK

Use the CUDA Compiler SDK to enable new languages to be GPU enabled , based on LLVM, this SDK provides all the essential files and documentation you need and  is now part of the CUDA Toolkit and not available as a separate download.

Now Part of CUDA Toolkit Install

Program Benefits

The CUDA Registered Developer Program is an important part of NVIDIA's commitment to the GPU Computing Developer community. It provides a mechanism for us to provide pre-release information and content to our most important developers and enabling two way communication about issues and feature requests.

If you have any suggestions on how to improve our developer program or have issues or problems with this program, please email us at CUDA_RegDev@nvidia.com