image and video datasets and models for torch deep learning

PyAMG is a library of Algebraic Multigrid (AMG) solvers with a convenient Python interface.

Implements iterative statistics operators for mean, variance, high-order moments, extrema, covariance, threshold, quantile (experimental) and Sobol' indices

This library provides ordinary differential equation (ODE) solvers implemented in PyTorch. Backpropagation through ODE solutions is supported using the adjoint method for constant memory cost. For usage of ODE solvers in deep learning applications.

As the solvers are implemented in PyTorch, algorithms in this repository are fully supported to run on the GPU.

Access dict values as attributes (works recursively).

Python implementation of the Tensor Train (TT) toolbox. It contains several important packages for working with the TT-format in Python. It is able to do TT-interpolation, solve linear systems, eigenproblems, solve dynamical problems. Several computational routines are done in Fortran (which can be used separately), and are wrapped with the f2py tool.

Example showing how to use starpu for implementing a distributed gemm in C++.

Chameleon is a dense linear algebra solver relying on sequential task-based algorithms where sub-tasks of the overall algorithms are submitted to a run-time system. Such a system is a layer between the application and the hardware which handles the scheduling and the effective execution of tasks on the processing units. A run-time system such as StarPU is able to manage automatically data transfers between not shared memory area (CPUs-GPUs, distributed nodes).

This software’s goal is to propose a parallel algebraic strategy to decompose a sparse linear system Ax=b, enabling its resolution by a domain decomposition solver. Up to now, Paddle is implemented for the MaPHyS linear solver.

PaRSEC is a generic framework for architecture aware scheduling and management of micro-tasks on distributed many-core heterogeneous architectures.

Mini-chameleon is an educational purpose dense linear algebra solver. As provided, it essentially provides drivers while the actual computational routines remain to be completed. The goal is to implement a dense matrix-matrix product and an LU factorization, first targeting a sequential implementation, followed by an simd version, a shared-memory openmp one, a distributed memory MPI one, an MPI+openmp one and a runtime-based starpu one.

Modified python 2.7.13. Scalable Python performs the I/O operations used e.g. by import statements in a single process and uses MPI to transmit data to/from all other processes.

QUARK (QUeuing And Runtime for Kernels) provides a library that enables the dynamic execution of tasks with data dependencies in a multi-core, multi-socket, shared-memory environment. QUARK infers data dependencies and precedence constraints between tasks from the way that the data is used, and then executes the tasks in an asynchronous, dynamic fashion in order to achieve a high utilization of the available resources.

ScalFMM is a C++ library that implements a kernel-independent Fast Multipole Method.

This is an open-source version of a simple application developed by Airbus for testing dense and sparse solvers with pseudo-FEM or pseudo-BEM matrices. Here in Guix, test_FEMBEM is currently connected to the open-source sequential version of hmat, Chameleon and H-Chameleon.

PaRSEC is a generic framework for architecture aware scheduling and management of micro-tasks on distributed many-core heterogeneous architectures.

The Basic Linear Algebra Subprograms (BLAS) have been around for many decades and serve as the de facto standard for performance-portable and numerically robust implementation of essential linear algebra functionality.The objective of BLAS++ is to provide a convenient, performance oriented API for development in the C++ language, that, for the most part, preserves established conventions, while, at the same time, takes advantages of modern C++ features, such as: namespaces, templates, exceptions, etc.

Chameleon is a dense linear algebra solver relying on sequential task-based algorithms where sub-tasks of the overall algorithms are submitted to a run-time system. Such a system is a layer between the application and the hardware which handles the scheduling and the effective execution of tasks on the processing units. A run-time system such as StarPU is able to manage automatically data transfers between not shared memory area (CPUs-GPUs, distributed nodes).

The Linear Algebra PACKage (LAPACK) is a standard software library for numerical linear algebra. The objective of LAPACK++ is to provide a convenient, performance oriented API for development in the C++ language, that, for the most part, preserves established conventions, while, at the same time, takes advantages of modern C++ features, such as: namespaces, templates, exceptions, etc.

Chameleon is a dense linear algebra solver relying on sequential task-based algorithms where sub-tasks of the overall algorithms are submitted to a run-time system. Such a system is a layer between the application and the hardware which handles the scheduling and the effective execution of tasks on the processing units. A run-time system such as StarPU is able to manage automatically data transfers between not shared memory area (CPUs-GPUs, distributed nodes).

pmtool aims at performing post-mortem analyses of the behavior of StarPU applications. Provide lower bounds on makespan. Study the performance of different schedulers in a simple context. Limitations: ignore communications for the moment; branch comms attempts to remove this limitation.

PaStiX (Parallel Sparse matriX package) is a scientific library that provides a high performance parallel solver for very large sparse linear systems based on direct methods. Numerical algorithms are implemented in single or double precision (real or complex) using LLt, LDLt and LU with static pivoting (for non symmetric matrices having a symmetric pattern). This solver also provides some low-rank compression methods to reduce the memory footprint and/or the time-to-solution.

PaStiX (Parallel Sparse matriX package) is a scientific library that provides a high performance parallel solver for very large sparse linear systems based on direct methods. Numerical algorithms are implemented in single or double precision (real or complex) using LLt, LDLt and LU with static pivoting (for non symmetric matrices having a symmetric pattern). This solver also provides some low-rank compression methods to reduce the memory footprint and/or the time-to-solution.