About lack of documentatio in some interfaces in MAGMA 1.3.0

Open discussion for MAGMA library (Matrix Algebra on GPU and Multicore Architectures)
Post Reply
Posts: 31
Joined: Tue Nov 15, 2011 12:38 pm

About lack of documentatio in some interfaces in MAGMA 1.3.0

Post by jgpallero » Fri Jan 11, 2013 7:00 am


I have some questions about the documentation in some MAGMA functions:

1. About magma_dgels_gpu:
Acoording to the interface

Code: Select all

double *dA,    magma_int_t ldda, 
double *dB,    magma_int_t lddb, 
double *hwork, magma_int_t lwork,
I understand that dA and dB matrices should be on the GPU and vector hwork in computer RAM (not in GPU). Am I right about the hvector?
The help in dgels_gpu.cpp says

Code: Select all

A       (input/output) DOUBLE_PRECISION array, dimension (LDA,N)
            On entry, the M-by-N matrix A.
            On exit, A is overwritten by details of its QR
            factorization as returned by DGEQRF.
so I think it should be updated in order to inform that the matrix should be on the GPU, as for the dB matrix says.

2. About magma_dgetri_gpu:

It has an error in the help. Where says:

Code: Select all

DWORK    (workspace/output) COMPLEX*16 array on the GPU, dimension (MAX(1,LWORK))
should say:

Code: Select all

DWORK    (workspace/output) DOUBLE PRECISION array on the GPU, dimension (MAX(1,LWORK))

Stan Tomov
Posts: 279
Joined: Fri Aug 21, 2009 10:39 pm

Re: About lack of documentatio in some interfaces in MAGMA 1

Post by Stan Tomov » Tue Feb 26, 2013 8:21 pm

Thanks for these documentation bug reports! We fixed them in the SVN.
Regarding the hvector question, you are right, it is on the CPU memory. An example on how to use the routine is in testing_dsgeqrsv_gpu.cpp. This routines tests mixed precision, as well as working precision solvers like the magma_dgels_gpu.

Post Reply