Package: blimps-examples Description-md5: 18ddc9ff8aab78bc1efd53cccc24389d Description-en: blocks database improved searcher (example data) BLIMPS (BLocks IMProved Searcher) is a searching tool that scores a protein sequence against blocks or a block against sequences. . This package contains example data. Package: blimps-utils Description-md5: 1afe4567b883b39e7050a4f604a91e17 Description-en: blocks database improved searcher BLIMPS (BLocks IMProved Searcher) is a searching tool that scores a protein sequence against blocks or a block against sequences. . This package contains the binaries. Package: cuda-cudart-13-1 Description-md5: 36541e97d3eee6603ad1d1302a21184c Description-en: CUDA Runtime native Libraries CUDA Runtime native Libraries Package: cuda-cudart-dev-13-1 Description-md5: 487ddbe32724f8f885d9d8cc19725975 Description-en: CUDA Runtime native dev links, headers CUDA Runtime native dev links, headers Package: cuda-driver-dev-13-1 Description-md5: fe2105d80e1ab1a9c25ca4dab8b47820 Description-en: CUDA Driver native dev stub library CUDA Driver native dev stub library Package: cuda-toolkit-13-1-config-common Description-md5: f360a76572469fb27116665b60e5bd1c Description-en: Common config package for CUDA Toolkit 13.1. Common config package for CUDA Toolkit 13.1. Handles /usr/local/cuda and /usr/local/cuda-13 alternatives. Package: cuda-toolkit-13-config-common Description-md5: 367821971543c6624e1066320f6f1b05 Description-en: Common config package for CUDA Toolkit 13. Common config package for CUDA Toolkit 13. Includes ld.so.conf config files. Package: cuda-toolkit-config-common Description-md5: 9d90ccaf5256a6008b4b6e334f608ae1 Description-en: Common config package for CUDA Toolkit. Common config package for CUDA Toolkit. Includes ld.so.conf config files. Package: dgx-desktop-app-profiles-mixed-coherency Description-md5: a8e07d3d719b355f93ad1b580085fa8a Description-en: NVIDIA Application Profiles as Mixed Coherency Systems NVIDIA GPU drivers provide an Application Profiles mechanism that allows system administrators to configure GPU selection policies and performance optimizations for specific applications based on their runtime characteristics. . In systems with both coherent (ATS-enabled) and non-coherent GPUs, certain applications may experience performance degradation or compatibility issues when automatically scheduled to ATS-capable devices. Application profiles enable fine-grained control over GPU assignment based on executable patterns, shared library dependencies, or process names. Package: dgx-desktop-arp-configuration Description-md5: 283cb8d8a3e08db52c9d53b79d119aae Description-en: Configure arp behavior for NVIDIA DGX Desktop Prevents routing ambiguity on systems with multiple network interfaces by ensuring all arp requests and responses use the correct network interface Package: dgx-desktop-cppc-cpufreq-options Description-md5: d914e4ae84f81a50ee1b2b2ee5477108 Description-en: cppc_cpufreq configuration specific to Grace Blackwell machines This package provides a cppc_cpufreq configuration specific to Grace Blackwell GB10 machines Package: dgx-desktop-crashkernel-configuration Description-md5: c57219e793679a81faab638cfbe19d34 Description-en: Configure crashkernel reservation for NVIDIA DGX Desktop Reserves no memory for crash kernel on systems with 1 GB or more of RAM. This configuration disables kdump crash dumping to maximize available memory for AI workloads. Package: dgx-desktop-disable-init-on-alloc Description-md5: 34aa50af725da3704fe0fcddc0bde7c1 Description-en: Disable init_on_alloc for NVIDIA DGX Desktop Prevents using CPU cycles to zero out memory before returning it to the caller. The requestor is the kernel, not user, thus already within the trust boundary and would be redundant. This is a huge perf impact if enabled on coherent memory systems. Package: dgx-desktop-disable-numa-balancing Description-md5: adacbdb06b23c7e7bc1bcc37ad780c4a Description-en: Disable numa balancing for NVIDIA DGX Desktop Disables automatic page fault NUMA memory balancing for a perf benefit on the Grace Blackwell systems. Package: dgx-desktop-docker-gpus Description-md5: e4892e60bb7c44e856acf87eec9f33f6 Description-en: Limit docker to 3D Controller class GPUs aka Compute GPUs This package limit nvidia docker to use 3D Controller class GPUs aka Compute GPUs. It is a systemd generator that creates a containerd service override file with the right DeviceAllow in it. Package: dgx-desktop-docker-options Description-md5: e943522749c0050dd9f66958f2bf49a2 Description-en: Docker daemon options Contains recommended docker daemon settings for NVIDIA systems. It creates an override configuration file for the docker service to force it to wait for the gpu-reset target before starting. Package: dgx-desktop-enable-persistenced Description-md5: 8e7e46abca092a904c3706a78ff89ceb Description-en: Enable NVIDIA GPU persistence mode NVIDIA GPUs can operate in persistence mode, which keeps the GPU driver loaded and initialized even when no applications are actively using the GPU. This reduces application startup latency and ensures consistent GPU state across workload lifecycles. Package: dgx-desktop-enable-power-meter-cap Description-md5: 7db9e884ae74623503f186e42c09ed5c Description-en: Enable power meter capability for NVIDIA DGX Desktop Exposes the power_cap sysfs nodes to enable finer grained power tuning. Package: dgx-desktop-kernel-configuration Description-md5: 17bec9f6714ca5344b8587de468ef9a7 Description-en: Kernel configuration for NVIDIA DGX Desktop This package provides kernel configuration parameters. It uses GRUB_FLAVOUR_ORDER in grub configuration to make sure that the -nvidia kernel will be the one prefered when booting the system. It also contains a systemd generator to load the right realtek driver according to the kernel used. Package: dgx-desktop-limits Description-md5: 02584d594ef173d975ccb769f47e9651 Description-en: Larger file descriptor limits to NVIDIA DGX Desktop High-performance GPU workloads often require increased file descriptor limits to support multiple concurrent processes and network connections Package: dgx-desktop-no-systemd-suspend Description-md5: 9aeb4fa067999cdbf5ef3bf34c965f85 Description-en: Prevents the machine to suspend Disables suspend using systemd drop-in configuration. GB10 devices are not compatible with suspend and this will make all suspend operations exit gracefully. Package: dgx-desktop-nvidia-fs-loader Description-md5: 97116c17ca30783e19aa565b639defac Description-en: Load the nvidia-fs module to enable GDS GPUDirect Storage (GDS) enables direct memory access (DMA) transfers between GPU memory and storage devices, bypassing CPU bounce buffers. This direct data path significantly increases system bandwidth, reduces latency, and decreases CPU utilization, making it essential for data-intensive AI and HPC workloads on the DGX Station GB300. Package: dgx-desktop-nvme-interrupt-coalescing Description-md5: 2745afe9a3744135bfd22b8d125f3f6a Description-en: Enable NVMe Interrupt coalescing on compatible systems Interrupt coalescing batches multiple I/O completion events into a single interrupt, improving system performance under high storage throughput workloads Package: dgx-desktop-pro-activation Description-md5: 9a5d0e88a2afda638dd4e89747012fe6 Description-en: Ubuntu Pro Activation for DGX Desktop Some DGX Desktop machines are shipped with Ubuntu Pro token preloaded. This package contains the necessary configuration to attach to the preloaded token and enable Ubuntu Pro. Package: dgx-desktop-sbsa-gwdt-options Description-md5: 2d6b1cbbb5676dde5a50e3056639849b Description-en: GRUB configuration scripts specific to NVIDIA GB10 machines It configures the watchdog accordingly to the vendor recommendations. It prevents a firmware race condition that would cause unattended reboots, before this upstream fix https://lkml.org/lkml/2025/7/8/1771 Package: firmware-qcom-dragonboard845c Description-md5: 3507088ed94312d44b4d5f5207d7374f Description-en: Binary firmware for various Qualcomm drivers used on Dragonboard 845c This package contains the binary firmware for GPU, USB, Venus, DSP hardware coprocessors found on SDM845, which is the main SoC on the Dragonboard 845c. Package: firmware-qcom-rb3gen2 Description-md5: 932dde057b2dbbe5130433652321aa51 Description-en: Binary firmware for Qualcomm Robotics RB3 Gen2 This package contains the binary firmware for the QCS6490, which is the main SoC on the Robotics RB3 Gen2. Package: firmware-qcom-rb5 Description-md5: 0abf8cbedf59bff7af0b4696550319c7 Description-en: Binary firmware for various Qualcomm drivers used on Robotics RB5 This package contains the binary firmware for the SM8250, which is the main SoC on the Robotics RB5. Package: gitaly-installer Description-md5: a0db51942a7c98ec821bffaf2b5a8789 Description-en: Git RPC service for handling all the git calls made by GitLab Gitaly makes the git data storage tier of large GitLab instances fast. This is achieved by moving git operations as close to the data as possible and Optimizing git services using caching. Gitaly is a core service of gitlab. This package installs Gitaly from pre-built binaries from Gitlab artifacts. Package: gitlab-common Description-md5: ccb3bd8dfffd14a6e75d53315368861a Description-en: git powered software platform to collaborate on code (common) gitlab provides web based interface to host source code and track issues. It allows anyone for fork a repository and send merge requests. Code review is possible using merge request workflow. Using groups and roles project access can be controlled. . This package includes configurations common to gitlab and gitaly. Package: hwe-dgx-gb10-meta Description-md5: 8c74d18bfb10e7484b87f70a6f1fc052 Description-en: Optimization Metapackage specific to NVIDIA DGX GB10 machines This package provide several system configuration to optimize and improve the performance of NVIDIA DGX GB10-based machines. Package: hwe-dgx-gb300ws-meta Description-md5: d5ed3c395933c930d480eec4dd2140e4 Description-en: Optimization Metapackage specific to NVIDIA DGX GB300ws machines This package provide several system configuration to optimize and improve the performance of NVIDIA DGX GB300 workstations. Package: libblimps3 Description-md5: a1ad50ab461eca726e0a4e957f1a2ffa Description-en: blocks database improved searcher library BLIMPS (BLocks IMProved Searcher) is a searching tool that scores a protein sequence against blocks or a block against sequences. . This package provides the shared library. Package: libblimps3-dev Description-md5: 4d04b155f279229d68533b0ec59a8e4f Description-en: blocks database improved searcher library (development) BLIMPS (BLocks IMProved Searcher) is a searching tool that scores a protein sequence against blocks or a block against sequences. . This package provides the library development headers and the static library. Package: libcublas-13-1 Description-md5: 4f31a4c3ca3e1747b91190d3816cbfac Description-en: CUBLAS native runtime libraries CUBLAS native runtime libraries Package: libcublas-dev-13-1 Description-md5: d8b51911296870ff057fbe1db1be45c2 Description-en: CUBLAS native dev links, headers CUBLAS native dev links, headers Package: libcurand-13-1 Description-md5: d77f8c451fd62a0c9c580f87176aa907 Description-en: CURAND native runtime libraries CURAND native runtime libraries Package: libcurand-dev-13-1 Description-md5: 39cc2c85fa90fd26bcbc4f0c6ebf162f Description-en: CURAND native dev links, headers CURAND native dev links, headers Package: libggml0-backend-cuda Description-md5: 56db844880e610d6c8401c92a13f05dd Description-en: Tensor library for machine learning - CUDA backend ggml is a tensor library for machine learning with the following features: . * Low-level cross-platform implementation * Integer quantization support * Broad hardware support * Automatic differentiation * ADAM and L-BFGS optimizers * No third-party dependencies * Zero memory allocations during runtime . On Debian, ggml is shipped as a package that depends on one of the CPU or GPU backends. . This package provides the CUDA backend, for NVIDIA GPUs. Package: libgloo-cuda-0 Description-md5: 11fe5b4bb9197845522ecac87176ae15 Description-en: Collective communications library (shared object) Gloo is a collective communications library. It comes with a number of collective algorithms useful for machine learning applications. These include a barrier, broadcast, and allreduce. . Transport of data between participating machines is abstracted so that IP can be used at all times, or InifiniBand (or RoCE) when available. In the latter case, if the InfiniBand transport is used, GPUDirect can be used to accelerate cross machine GPU-to-GPU memory transfers. . Where applicable, algorithms have an implementation that works with system memory buffers, and one that works with NVIDIA GPU memory buffers. In the latter case, it is not necessary to copy memory between host and device; this is taken care of by the algorithm implementations. . This package ships the shared object for Gloo. Package: libgloo-cuda-dev Description-md5: 1eb5d67397fdb9a8162c5c082d1e6d78 Description-en: Collective communications library (development files) Gloo is a collective communications library. It comes with a number of collective algorithms useful for machine learning applications. These include a barrier, broadcast, and allreduce. . Transport of data between participating machines is abstracted so that IP can be used at all times, or InifiniBand (or RoCE) when available. In the latter case, if the InfiniBand transport is used, GPUDirect can be used to accelerate cross machine GPU-to-GPU memory transfers. . Where applicable, algorithms have an implementation that works with system memory buffers, and one that works with NVIDIA GPU memory buffers. In the latter case, it is not necessary to copy memory between host and device; this is taken care of by the algorithm implementations. . This package ships the development files. Package: libnvjitlink-13-1 Description-md5: 44a558ba3d0f0d9876336d32eaa51aab Description-en: JIT linking for NVIDIA GPU device code This package contains the runtime libraries necessary to just-in-time link NVIDIA GPU device code. It also supports Link Time Optimization. Package: libnvjitlink-dev-13-1 Description-md5: 5b8ac34b251131806612e86213ce030f Description-en: Development files for libnvjitlink This package contains the headers and static library files necessary for building C programs which use libnvjitlink. Package: libparmetis-dev Description-md5: 839c770f477cb92f6af09275d807c484 Description-en: Parallel Graph Partitioning and Sparse Matrix Ordering Libs: Devel ParMetis computes minimal-cut partitions of graphs and meshes in parallel, and orders variables for minimal fill when using direct solvers for sparse matrices. It does all this in parallel, and also can efficiently re-partition a graph or mesh whose connectivity has changed. . This package contains files needed to develop programs using ParMetis. Package: libparmetis4.0 Description-md5: 17a6686f47a3b63f4328881bffab697b Description-en: Parallel Graph Partitioning and Sparse Matrix Ordering Shared Libs ParMetis computes minimal-cut partitions of graphs and meshes in parallel, and orders variables for minimal fill when using direct solvers for sparse matrices. It does all this in parallel, and also can efficiently re-partition a graph or mesh whose connectivity has changed. . This package contains the ParMetis shared libraries. Package: libtensorpipe-cuda-0 Description-md5: b49a93726d4da68e6161824dc04f12b3 Description-en: tensor-aware point-to-point communication primitive for machine learning The TensorPipe project provides a tensor-aware channel to transfer rich objects from one process to another while using the fastest transport for the tensors contained therein (e.g., CUDA device-to-device copy). . This package contains shared objects. Package: libtensorpipe-cuda-dev Description-md5: 6b4397dd30e29fe70ebb4ea9c71ececf Description-en: tensor-aware point-to-point communication primitive for machine learning The TensorPipe project provides a tensor-aware channel to transfer rich objects from one process to another while using the fastest transport for the tensors contained therein (e.g., CUDA device-to-device copy). . This package contains development files. Package: parmetis-doc Description-md5: 55cc39b179c0b5b2dedead6fc6a8c34f Description-en: Parallel Graph Partitioning and Sparse Matrix Ordering Lib - Docs ParMetis computes minimal-cut partitions of graphs and meshes in parallel, and orders variables for minimal fill when using direct solvers for sparse matrices. It does all this in parallel, and also can efficiently re-partition a graph or mesh whose connectivity has changed. . This package contains the documentation and example files. Package: parmetis-test Description-md5: b9b53f52a3b7e53d03b5260911e600a9 Description-en: Parallel Graph Partitioning and Sparse Matrix Ordering Tests ParMetis computes minimal-cut partitions of graphs and meshes in parallel, and orders variables for minimal fill when using direct solvers for sparse matrices. It does all this in parallel, and also can efficiently re-partition a graph or mesh whose connectivity has changed. . This package contains programs which test the ParMetis libraries using files in the parmetis-doc package's examples directory. Package: python-cx-oracle-doc Description-md5: 590c60b950dbdb54858ff4c6e9650780 Description-en: Python interface to Oracle Database (common documentation) cx_Oracle is a Python extension module that enables access to Oracle Database. . It conforms to the Python database API 2.0 specification with a considerable number of additions and a couple of exclusions. . You can use cx_Oracle with Oracle 11.2, 12, 18 and 19 client libraries. . This is the common documentation package. Package: python3-cx-oracle Description-md5: ea1583d4bbdab3ed20ac652efb13bda1 Description-en: Python interface to Oracle Database (Python 3) cx_Oracle is a Python extension module that enables access to Oracle Database. . It conforms to the Python database API 2.0 specification with a considerable number of additions and a couple of exclusions. . You can use cx_Oracle with Oracle 11.2, 12, 18 and 19 client libraries. . This package installs the library for Python 3. Package: sift Description-md5: 7788bf12148938f8dc8e4675657ce605 Description-en: predicts if a substitution in a protein has a phenotypic effect SIFT is a sequence homology-based tool that sorts intolerant from tolerant amino acid substitutions and predicts whether an amino acid substitution in a protein will have a phenotypic effect. SIFT is based on the premise that protein evolution is correlated with protein function. Positions important for function should be conserved in an alignment of the protein family, whereas unimportant positions should appear diverse in an alignment. Package: libodpic5 Description-md5: ab5d684e6cf00e24207e0ddef1bfe15f Description-en: Oracle Database Programming Interface - runtime C shared library Oracle Database Programming Interface for C (ODPI-C) is an open source library of C code that simplifies access to Oracle Database for applications written in C or C++. It is a wrapper over Oracle Call Interface (OCI) that makes applications and language interfaces easier to develop. . ODPI-C supports basic and advanced features of Oracle Database and Oracle Client. See the homepage for a list. . This package contains the C shared library needed to run programs that dynamically link to ODPI-C. Package: odpic-dev Description-md5: 49a15361421aff7972b2ab791fe2768e Description-en: Oracle Database Programming Interface - development files Oracle Database Programming Interface for C (ODPI-C) is an open source library of C code that simplifies access to Oracle Database for applications written in C or C++. It is a wrapper over Oracle Call Interface (OCI) that makes applications and language interfaces easier to develop. . ODPI-C supports basic and advanced features of Oracle Database and Oracle Client. See the homepage for a list. . This package contains the files needed to develop programs that use ODPI-C; including the headers, examples, and sample programs.