https://github.com/dthuerck/culip

Code for the culip ("CUda for Linear and Integer Programming") project, containing GPU primitives for linear algebra, linear optimization and (someday) integer optimization.
https://github.com/dthuerck/culip

Last synced: 27 days ago
JSON representation

Code for the culip ("CUda for Linear and Integer Programming") project, containing GPU primitives for linear algebra, linear optimization and (someday) integer optimization.

• Host: GitHub
• URL: https://github.com/dthuerck/culip
• Owner: dthuerck
• License: other
• Created: 2018-10-08T07:17:59.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2018-10-12T19:01:00.000Z (over 6 years ago)
• Last Synced: 2025-04-01T12:50:51.606Z (about 2 months ago)
• Language: Cuda
• Size: 14 MB
• Stars: 18
• Watchers: 4
• Forks: 3
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# culip

![culip logo](doc/logo.png)

## Project description

The **cu**da for **l**inear and **i**nteger **p**rogramming project contains

a collection of GPU primitives for

* linear algebra,

* linear programming and

* integer programming.

## Source directory structure

* apps - Executables using the modules in lib/

* dependencies - Building scripts for external dependencies

* libs - This is where the magic happens - all functional/backend code pieces

  * algorithms - symbolical/discrete CPU algorithms

  * data_structures - specialized, templated CPU data structures (ixheap, kvheap, ...)

  * la - Linear algebra modules (wrapper, kkt adapters, iterative solvers, preconditioners)

  * util - various utilities, such as a memory manager

* tests - Tests for executables and libs

## Dependencies

* [NVIDIA Thrust](https://developer.nvidia.com/thrust)

* [NVIDIA CUB](https://nvlabs.github.io/cub/)

* [MMIO](https://math.nist.gov/MatrixMarket/)

* [CBlas](https://www.netlib.org/blas/)

* [LAPACKE](http://www.netlib.org/lapack/lapacke.html)

* [OpenBLAS](https://www.openblas.net/)

All dependencies are downloaded and built automatically in the provided CMake build process.

## Get started

The build process is based on CMake and has been tested on Ubuntu 18.04 with gcc 7.3.0 and CUDA >= 9.2.

Just execute the following after cloning and cd'ing into the cloned folder:

```

$ mkdir build

$ cd build

$ cmake ..

$ ccmake .

$ make

```

In the GUI appearing after `ccmake`, select the parts you wish to build and enter the `CUDA_SDK_ROOT_DIR`, which

should point to the base dir of NVIDIA's CUDA samples included in their CUDA distribution.

## License and Copyright

Copyright (c) 2016, NVIDIA CORPORATION. All rights reserved. 


Copyright (c) 2017, Daniel Thuerck, TU Darmstadt - GCC. All rights reserved.

This software may be modified and distributed under the terms

of the BSD 3-clause license. See the LICENSE file for details.