User Tools

Site Tools


faq:cuda_support

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
faq:cuda_support [2014/04/06 14:32]
vondele
faq:cuda_support [2015/02/18 15:55] (current)
oschuett
Line 1: Line 1:
 ====== Which parts of CP2K are CUDA-accelerated?​ ====== ====== Which parts of CP2K are CUDA-accelerated?​ ======
  
-  * Anything that uses ''​dbcsr_multiply'',​ i.e. sparse matrix multiplication,​ when compiled with ''​%%-D__DBCSR_CUDA%%''​. This benefits in particular the [[doi>​10.1021/​ct200897x| linear scaling DFT]] code. See also [[http://​dbcsr.cp2k.org ​the DBCSR project.]] +Have a look at [[howto:compile_with_cuda ​this howto ]].
-  * If linked against an accelerated scalapack/​blas library (in particular pdgemm/​pdsyrk/​dgemm) that executes these calls on the GPU. The impact of this is most visible for MP2 and RPA calculations. On the hybrid Cray XC30 linking against libsci_acc makes this happen. +
-  * FFTs, when compiled with ''​%%-D__PW_CUDA%%''​. +
- +
  
  
faq/cuda_support.txt · Last modified: 2015/02/18 15:55 by oschuett