You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have encountered some issues while using SuperLU_DIST and would appreciate your assistance.
When using the command "mpiexec -n 8 pddrive myrhs mymatrix", I obtain solutions with very small residuals. However, when I use "mpiexec -n 8 pddrive -r 2 -c 4 myrhs mymatrix", the residuals become significantly larger. I tried using the -i 1 and -i 2 options, but they did not improve the situation.
Additionally, I am working with matrices that have up to 10 million rows. In this context, should I use the -DXSDK_INDEX_SIZE=64 option?
Thank you for your assistance.
Best regards,
[Muxing Li]
The text was updated successfully, but these errors were encountered:
If it succeeds with 1 MPI 32-bit indexing, you don't need 64 bit for multiple MPIs.
Can you compile a version with more printings, by defining:
-DCMAKE_C_FLAGS="-DPRNTlevel=1"
Dear SuperLU Team,
Hello!
I have encountered some issues while using SuperLU_DIST and would appreciate your assistance.
When using the command "mpiexec -n 8 pddrive myrhs mymatrix", I obtain solutions with very small residuals. However, when I use "mpiexec -n 8 pddrive -r 2 -c 4 myrhs mymatrix", the residuals become significantly larger. I tried using the -i 1 and -i 2 options, but they did not improve the situation.
Additionally, I am working with matrices that have up to 10 million rows. In this context, should I use the -DXSDK_INDEX_SIZE=64 option?
Thank you for your assistance.
Best regards,
[Muxing Li]
The text was updated successfully, but these errors were encountered: