3 CUDA-only packages are stored in the `cudaPackages` packages set. This set
4 includes the `cudatoolkit`, portions of the toolkit in separate derivations,
5 `cudnn`, `cutensor` and `nccl`.
7 A package set is available for each CUDA version, so for example
8 `cudaPackages_11_6`. Within each set is a matching version of the above listed
9 packages. Additionally, other versions of the packages that are packaged and
10 compatible are available as well. For example, there can be a
11 `cudaPackages.cudnn_8_3` package.
13 To use one or more CUDA packages in an expression, give the expression a `cudaPackages` parameter, and in case CUDA is optional
16 , cudaSupport ? config.cudaSupport
22 When using `callPackage`, you can choose to pass in a different variant, e.g.
23 when a different version of the toolkit suffices
25 mypkg = callPackage { cudaPackages = cudaPackages_11_5; }
28 If another version of say `cudnn` or `cutensor` is needed, you can override the
29 package set to make it the default. This guarantees you get a consistent package
33 cudaPackages = cudaPackages_11_5.overrideScope (final: prev: {
34 cudnn = prev.cudnn_8_3;
36 in callPackage { inherit cudaPackages; };
39 The CUDA NVCC compiler requires flags to determine which hardware you
40 want to target for in terms of SASS (real hardware) or PTX (JIT kernels).
42 Nixpkgs tries to target support real architecture defaults based on the
43 CUDA toolkit version with PTX support for future hardware. Experienced
44 users may optimize this configuration for a variety of reasons such as
45 reducing binary size and compile time, supporting legacy hardware, or
46 optimizing for specific hardware.
48 You may provide capabilities to add support or reduce binary size through
49 `config` using `cudaCapabilities = [ "6.0" "7.0" ];` and
50 `cudaForwardCompat = true;` if you want PTX support for future hardware.
52 Please consult [GPUs supported](https://en.wikipedia.org/wiki/CUDA#GPUs_supported)
53 for your specific card(s).
55 Library maintainers should consult [NVCC Docs](https://docs.nvidia.com/cuda/cuda-compiler-driver-nvcc/)
56 and release notes for their software package.
58 ## Adding a new CUDA release {#adding-a-new-cuda-release}
62 > This section of the docs is still very much in progress. Feedback is welcome in GitHub Issues tagging @NixOS/cuda-maintainers or on [Matrix](https://matrix.to/#/#cuda:nixos.org).
64 The CUDA Toolkit is a suite of CUDA libraries and software meant to provide a development environment for CUDA-accelerated applications. Until the release of CUDA 11.4, NVIDIA had only made the CUDA Toolkit available as a multi-gigabyte runfile installer, which we provide through the [`cudaPackages.cudatoolkit`](https://search.nixos.org/packages?channel=unstable&type=packages&query=cudaPackages.cudatoolkit) attribute. From CUDA 11.4 and onwards, NVIDIA has also provided CUDA redistributables (“CUDA-redist”): individually packaged CUDA Toolkit components meant to facilitate redistribution and inclusion in downstream projects. These packages are available in the [`cudaPackages`](https://search.nixos.org/packages?channel=unstable&type=packages&query=cudaPackages) package set.
66 All new projects should use the CUDA redistributables available in [`cudaPackages`](https://search.nixos.org/packages?channel=unstable&type=packages&query=cudaPackages) in place of [`cudaPackages.cudatoolkit`](https://search.nixos.org/packages?channel=unstable&type=packages&query=cudaPackages.cudatoolkit), as they are much easier to maintain and update.
68 ### Updating CUDA redistributables {#updating-cuda-redistributables}
70 1. Go to NVIDIA's index of CUDA redistributables: <https://developer.download.nvidia.com/compute/cuda/redist/>
71 2. Copy the `redistrib_*.json` corresponding to the release to `pkgs/development/compilers/cudatoolkit/redist/manifests`.
72 3. Generate the `redistrib_features_*.json` file by running:
75 nix run github:ConnorBaker/cuda-redist-find-features -- <path to manifest>
78 That command will generate the `redistrib_features_*.json` file in the same directory as the manifest.
80 4. Include the path to the new manifest in `pkgs/development/compilers/cudatoolkit/redist/extension.nix`.
82 ### Updating the CUDA Toolkit runfile installer {#updating-the-cuda-toolkit}
86 > While the CUDA Toolkit runfile installer is still available in Nixpkgs as the [`cudaPackages.cudatoolkit`](https://search.nixos.org/packages?channel=unstable&type=packages&query=cudaPackages.cudatoolkit) attribute, its use is not recommended and should it be considered deprecated. Please migrate to the CUDA redistributables provided by the [`cudaPackages`](https://search.nixos.org/packages?channel=unstable&type=packages&query=cudaPackages) package set.
88 > To ensure packages relying on the CUDA Toolkit runfile installer continue to build, it will continue to be updated until a migration path is available.
90 1. Go to NVIDIA's CUDA Toolkit runfile installer download page: <https://developer.nvidia.com/cuda-downloads>
91 2. Select the appropriate OS, architecture, distribution, and version, and installer type.
93 - For example: Linux, x86_64, Ubuntu, 22.04, runfile (local)
94 - NOTE: Typically, we use the Ubuntu runfile. It is unclear if the runfile for other distributions will work.
96 3. Take the link provided by the installer instructions on the webpage after selecting the installer type and get its hash by running:
99 nix store prefetch-file --hash-type sha256 <link>
102 4. Update `pkgs/development/compilers/cudatoolkit/versions.toml` to include the release.
104 ### Updating the CUDA package set {#updating-the-cuda-package-set}
106 1. Include a new `cudaPackages_<major>_<minor>` package set in `pkgs/top-level/all-packages.nix`.
108 - NOTE: Changing the default CUDA package set should occur in a separate PR, allowing time for additional testing.
110 2. Successfully build the closure of the new package set, updating `pkgs/development/compilers/cudatoolkit/redist/overrides.nix` as needed. Below are some common failures:
112 | Unable to ... | During ... | Reason | Solution | Note |
113 | --- | --- | --- | --- | --- |
114 | Find headers | `configurePhase` or `buildPhase` | Missing dependency on a `dev` output | Add the missing dependency | The `dev` output typically contain the headers |
115 | Find libraries | `configurePhase` | Missing dependency on a `dev` output | Add the missing dependency | The `dev` output typically contain CMake configuration files |
116 | Find libraries | `buildPhase` or `patchelf` | Missing dependency on a `lib` or `static` output | Add the missing dependency | The `lib` or `static` output typically contain the libraries |
118 In the scenario you are unable to run the resulting binary: this is arguably the most complicated as it could be any combination of the previous reasons. This type of failure typically occurs when a library attempts to load or open a library it depends on that it does not declare in its `DT_NEEDED` section. As a first step, ensure that dependencies are patched with [`cudaPackages.autoAddOpenGLRunpath`](https://search.nixos.org/packages?channel=unstable&type=packages&query=cudaPackages.autoAddOpenGLRunpath). Failing that, try running the application with [`nixGL`](https://github.com/guibou/nixGL) or a similar wrapper tool. If that works, it likely means that the application is attempting to load a library that is not in the `RPATH` or `RUNPATH` of the binary.