Stack Overflow
Stack Overflow

Questions tagged [cusolver]

a library that is part of the NVIDIA CUDA toolkit to accelerate various dense and sparse LAPACK-like operations on GPUs

cusolver is a library included with CUDA 7.0 and beyond, to provide GPU acceleration for 3 categories of linear algebra problems:

  1. cuSolverDN deals with dense matrix factorization and solve routines such as LU, QR, SVD and LDLT

  2. cuSolverSP provides a set of sparse routines based on a sparse QR factorization

  3. cuSolverRF is a sparse re-factorization package for solving a sequence of matrices where only the coefficients are changed but the sparsity pattern remains the same

cusolver documentation is here.

56 questions
0
votes
1 answer

Getting incorrect result with cusolverDnDpotrfBatched

I want to find the Cholesky decomposition of a 3x3 matrix using cusolverDnDpotrfBatched, but I am not getting the zeros that should be present in a lower triangular matrix. Here is the matrix for which I want to compute the cholesky decomposition [1…
Anik Chaudhuri
  • 65
  • 1
  • 6
0
votes
0 answers

Why are eigenvectors computed only every other function call in Cusolver?

I'm writing a program to compare the speed of JBlas and JCublas. When I call the below function the first time, everything works fine and v contains the correct eigenvectors. When I call it a second time, it takes a lot less time to compute but only…
BluSteve
  • 1
0
votes
1 answer

wrong results from cusolverDnDDgels

I wanted to test the cuda implementation of xgels provided with CUDA 11.1, and it seems I cannot make it work properly. For instance, this code seems to run just fine: #include #include int main() { const int m =…
Enzo Ferrazzano
  • 157
  • 10
0
votes
0 answers

How to find eigenvalues of a big matrix in parallel?

I need to find eigenvalue of a matrices 1000 on 1000 and bigger with CUDA in parallel as fast as possible. I found cusolver library and run the code from documentation: #include #include #include #include…
anonymus321
  • 33
  • 4
0
votes
1 answer

output results mismatch, LAPACKE_zheev() vs cusolverDnZheevd()

I am translating code from legacy cblas/lapacke to cudaBLAS/cudaSOLVER and having some issues. I made a test program to get into the bottom of this. Attached is the code that I am using: #include #include #include…
carbuet
  • 1
  • 1
0
votes
1 answer

LU factorization receives different results between LAPACK and cuBLAS/cuSOLVER

I am testing out some scenarios where the function dgetrf is returned differently when used with cuBLAS/cuSOLVER compared to writing for LAPACK. For example, I am looking at LU factorization of the following matrix: 2.0 4.0 1.0 -3.0 0.0 -1.0 -2.0…
Quang Thinh Ha
  • 239
  • 2
  • 10
0
votes
1 answer

CUDA cuSolver gesvdj with large matrix

I am running a slightly modified version of the code found here under section "G.2. SVD with singular vectors (via Jacobi method)" on a NVIDIA P6000. The slight modifications are to dynamically allocate memory in the heap for the A, U, and V vectors…
Will.Evo
  • 1,112
  • 13
  • 31
0
votes
1 answer

cusparseSdense2csr conversion

I am trying to convert a dense matrix to sparse using the cusparseSdense2csr API, the dense matrix is as follows : [ 0 1 0 3 0 0 6 0 8 0 0 11 0 13 0 0 16 0 18 0 ] The expected resultant sparse matrix should be : csrValA =…
Prz9
  • 33
  • 7
0
votes
1 answer

Does cuSolverDN or another CUDA library have a batched-version of QR decomposition for dense matrices to solve A*x = b?

I'm trying to solve A*x = b where A has complex values and is dense. I used cusolverDnCgeqrf() method from cuSolverDN library to do the QR decomposition for one linear set of equations. However, I want to do this several times to speed up the…
David K
  • 124
  • 4
0
votes
1 answer

Proper use of cudaFortran cuSolver functions

I am currently working on migrating some Fortran code over to cudaFortran. Specifically the task involves the spectral analysis of massive matrices in order to diagonalize them. Here's the code I've fabricobbled so far program main !Trials for usage…
Ólafur
  • 1
  • 5
0
votes
1 answer

What version of CUDA introduced cuSolverRF?

I have access to a machine with CUDA 7.5, but I think it is missing the cuSolverRF library. The current documentation marked as version DU-06709-001_v8.0 says "In addition cuSolver provides a new refactorization library useful for solving sequences…
Peter
  • 1,312
  • 3
  • 15
  • 16
0
votes
1 answer

Compiling my CUDA program with libraries provided in toolkit

I wrote simple CUDA c++ program simulating diffusion on 2D matrix. I got in trouble when I tried to used some of the libraries which are provided in Toolkit. I would like to replace my extremely inefficient matrix transpose kernel with something…
Max K
  • 71
  • 1
  • 11
0
votes
1 answer

CuSolver Sparse interface on Fortran

I'm trying to write a program to interface cusolverSp onto fortran. although I'm no stranger for coding cuda in C, I'm unsure how to get it on fortran. The following is my code: ! Fortran Console Application ! module cuda_cusolverSP …
ceeely
  • 135
  • 1
  • 10
0
votes
1 answer

huge use of cudaFree from cuSparse tridiagonal solver

I am using cusparseDgtsv_nopivot function to solve a tridiagonal system of equation. the output is correct but the function does not make proper use of cuda multi-streaming. The nvvp profiler shows that although every call to this solver is in a…
Med Aissa
  • 3
  • 2
0
votes
2 answers

Multiple definition errors of cuSparse functions while linking CUDA files in an executable

My project consists of five CUDA files: main.cu jacobian_kernel.cu hermite_kernel.cu cuSolver_LU.cpp Utilities.cu, the last of which is adapted from this GitHub repo, together with its Utilities.h header file; the three headers are args.h…
Eugenio
  • 244
  • 2
  • 13
1 2
3
4