Update streams correspond to either all precompiled driver updates on specific driver branches (e.g., drivers with major version equal to "440"). Only updates on the selected stream will be considered.
Using Modularity, the tech preview repository provides multiple update streams of the driver packages. We expect any potential delays to be short. This can delay the application of security fixes, but ensures that a tested kernel/driver combination is used at all times.
When using precompiled drivers, a dnf plugin is enabled that prevents upgrading to a kernel for which no precompiled driver exists yet (a warning will be displayed by dnf during such an upgrade situation). In some cases such as embargoed kernel updates, precompiled driver packages for a kernel may not be publicly available before the availability of this kernel. Note that "most recent kernel" includes all RHEL z-stream updates. If using an older kernel release, the matching older precompiled driver release can be used. Thus, this is the same constraint as for other kernel modules that are shipped as part of RHEL. Another way to phrase this is that at any point in time, precompiled drivers are enabled for the most recent kernel and the most recent driver at this point in time. When a new driver update is released, precompiled driver packages are provided only for the most recently released kernel at the time of the driver update likewise, if a new kernel update is released, precompiled driver packages will be provided for this kernel. The source files for these driver kmod packages are compiled in advance and then linked at installation time, hence these are called "precompiled drivers". The new approach does not require gcc to be installed anymore, nor does the EPEL repository need to be enabled.
This in turn means that I need to install an nVidia driver that is compatible with my CUDA version. For instance, my laptop has an nVidia CUDA 2.1 GPU, which means I can't install a CUDA toolkit more recent than CUDA 8.0 GA2.
A GPU will support a specific compute version, and if your CUDA version minimum compute capability is higher than that supported by your GPU, your code will not compile or run. The other half is the Compute Capability. The driver version is only half of the compatibility equation. The general support matrix is contained in the linux install guide for the specific CUDA version of interest.įor reference, on linux, the previous CUDA toolkits required the following minimum driver versions: For versions newer than 11.0, refer to the table in the release notes linked belowĪ portion of this table is maintained as Table 1 here in the release notes. Not all distros are supported on every CUDA toolkit version. Note that any given CUDA toolkit has specific Linux distros (including version number) that are supported. Older CUDA toolkits are available for download here. Starting with CUDA 9.x, older CUDA GPUs of compute capability 2.x are also not supported. Note that CUDA 7 will not be usable with older CUDA GPUs of compute capability 1.x. This is not necessary if you want to use CUDA 7, you can just use the runfile installer method which has all the necessary components including an appropriate driver (346.46). I believe you are picking up a 304.xx driver via a specific (ie. If you want to reinstall ubuntu to create a clean setup, the linux getting started guide has all the instructions needed to set up CUDA if that is your intent. 304.xx is a driver that will support CUDA 5 and previous (does not support newer CUDA versions.)