Skip to content

Developing CUDA kernels to accelerate the list search in the IMPALA algorithm.

License

Notifications You must be signed in to change notification settings

oli-dj/impala-gpu

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

impala-gpu

MPS algorith using the principles of SNESIM / IMPALA, implemented in MATLAB with core parts accellerated by GPU. GPU implementation can handle conditioning to hard data, soft data (both colocational and not). CPU implementation does not handle non-colocational soft data yet.

Developed in MATLAB R2017b, not tested in earlier versions. GPU code requires CUDA installed and is tested to work with NVIDIA Pascal or compatible graphicscards. Earlier cards may work. Most cards will work, but may require compiling of the .cu files for their architiceture.

Requires https://github.com/cultpenguin/mGstat in path for mps_template.m and channels.m.

Installation

Clone and use directly from MATLAB, see example_script.m for an example.

Set "options.GPU = 0;" if no CUDA capable GPU avaliable, note: will not handle non-colocational soft data in this case.

About

Developing CUDA kernels to accelerate the list search in the IMPALA algorithm.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published