Since cuDNN is split into several libraries, dependencies between them need to be taken into account.įor example, when statically linking libcudnn_cnn_infer_static.a into an application, libcudnn_ops_infer_static.a is also needed, in this order (a dependent library followed by its dependency). Static cuDNN libs for Windows are not supported. Linux: Add -lcublas_static -lcublasLt_static -lz -lculibos -lnvrtc_static -lnvrtc-builtins_static -lnvptxcompiler_static -lcudart_static to the linker command. Linker dependencies for the static cuDNN libs Windows: Add cublas.lib cublasLt.lib zlibwapi.lib to the linker command. Linux: Add -lcublas -lcublasLt -lz to the linker command. One way to achieve this is by explicitly specifying them on the linker command.įor linker dependencies for the dynamic cuDNN libs Navigate to your directory containing the cuDNN tar file.īecause cuDNN uses symbols defined in external libraries, you need to ensure that the linker can locate these libraries while building a cuDNN dependent program.your cuDNN download path is referred to as īefore issuing the following commands, you must replace X.Y and v8.x.x.x with your specific CUDA and cuDNN versions and package date.your CUDA directory path is referred to as /usr/local/cuda/.The RPM package installation applies to RHEL7, RHEL8, and RHEL9. The Debian package installation applies to Debian 11, Ubuntu 18.04, Ubuntu 20.04, and 22.04. For example, the tar file installation applies to all Linux platforms. Choose the installation method that meets your environment needs. The following steps describe how to build a cuDNN dependent program. Select the cuDNN version that you want to install.A list of available download versions of cuDNN displays. Complete the short survey and click Submit.In order to download cuDNN, ensure you are registered for the NVIDIA Developer Program.
0 kommentar(er)
