Regarding dense matrices and 1D distribution scheme

Post here if you have a question about LAPACK or ScaLAPACK algorithm or data format

Regarding dense matrices and 1D distribution scheme

Postby anasaltrad » Tue Sep 13, 2011 3:20 am

Hi, I am newbie to ScaLAPACK and I am writing a simple program which will calculate the matrix multiplication on a Linux Cluster. I've read the ScaLAPACK standard and it says that:

The two-dimensional block cyclic distribution scheme is the data layout that is used in the ScaLAPACK library for dense matrix computations.

Does this imply that I cannot use 1D block cyclic distribution scheme for dense matrices? In other words I want to compute C = A*B where A is distributed in row-wise block cyclic, B in column-wise block cyclic, and then call pdgemm_ function. Does such an algorithm comply to ScaLAPACK standard and ScaLAPACK will give the correct answer?

Thanks in advance
anasaltrad
 
Posts: 3
Joined: Tue Sep 13, 2011 2:57 am

Return to Algorithm / Data

Who is online

Users browsing this forum: Baidu [Spider] and 1 guest