Failure of Eigenvalue Decomposition for Large Matrices. #2413
Labels
bug
Something isn't working
cuda libraries
Stuff about CUDA library wrappers.
good first issue
Good for newcomers
Describe the bug
Before updating, I was able to decompose a large (100,000 by 100,000) dense matrix using the CUDA.CUSOLVER.syevd! with CUDA.jl version 5.2.0. However, after updating to version 5.4.2, the eigenvalue decomposition (CUDA.CUSOLVER.syevd!) fails even with a 15,000 by 15,000 dense matrix, resulting in the following error message:
To reproduce
The Minimal Working Example (MWE) for this bug:
Manifest.toml
Version info
Details on Julia:
Details on CUDA:
The text was updated successfully, but these errors were encountered: