Open fxuecnu opened 7 months ago
Hello! Can you provide a matrix itself? I'm collecting data for extreme test cases for my solver and that would be a great help! Plus, i might be able to give some advice on the configuration of AMGX if i have the matrix.
@evstigneevnm Hello! Thank you for getting back to me.
I've attached the matrix information as requested. If you encounter any issues or have further questions, please don't hesitate to reach out.
Hello,
I'm working on solving a matrix equation that seems to be ill-conditioned, and I'm encountering convergence issues with all the configuration files provided in AMGX. I've attached a JPEG image of the matrix to give an idea of its structure.
Despite confirming that the matrix rank is appropriate, I'm struggling with the equation's principal axis, which contains many zero values. This characteristic might be contributing to the convergence problems.
I'm seeking advice on how to properly configure AMGX for this type of equation. To provide more context, I've also attached a snippet of my code that illustrates how I'm constructing and attempting to solve this system.
Any suggestions or guidance on the appropriate AMGX configuration settings for this scenario would be greatly appreciated.
Thank you for your time and help.
//// below is code
/// below is the message returned
C:\STUDY\testamgx\CudaRuntime1\x64\Release>CudaRuntime1.exe -p 3 3 -c C:\AMGX\AMGX-2.3.0\core\configs\FGMRES_NOPREC.json Process 0 selecting device 0 AMGX version 2.2.0.132-opensource Built on Nov 23 2023, 14:14:00 Compiled with CUDA Runtime 11.8, using CUDA driver 12.2 Warning: No mode specified, using dDDI by default. Cannot read file as JSON object, trying as AMGX config Converting config string to current config version Parsing configuration string: exception_handling=1 ; idx = 0, nnz = 0 Using Normal MPI (Hostbuffer) communicator... iter Mem Usage (GB) residual rate
Total Time: 0.0932096 setup: 8.192e-05 s solve: 0.0931277 s solve(per iteration): 0.000931277 s
/// below is the shape of the sparse matrix
///