Skip to content

ravikanthreddy89/cuda

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

LU decomposition using CUDA

  1. A parallel implementation of LU decomposition

  2. Two versions : 1) Using global memory alone 2) Using shared memory for pivot row

  3. For both the implementations kernel with single thread scales the pivot row

  4. Global memory : Blocks with one thread each are launched for reduction.

  5. Shared memory : Blocks with static size are used. Thread with id==0 copies the pivot row into shared memory and after that rest of the threads in the block start reducing.

About

LU Decomposition using CUDA

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages