I am using the Cluster Sparse Solver on Windows 64 (Parallel Studio xe 2016.0.110) on a symmetric indefinite matrix and the numeric factorization crashes for 2 MPI process and 4 OpenMP threads. No specific error message is displayed.
In case of 1 or 4 MPI processes with 4 OpenMP threads the calculation is running without crash.
My C++ program is compiled with the Microsoft Visual Studio C++ compiler (Visual Studio 12 2013 18.0.31101.0) and is linked with the following libraries:
- impi.lib (Intel MPI 126.96.36.199 of Parallel Studio xe 2016.0.110)
- mkl_intel_lp64.lib;mkl_core.lib;mkl_intel_thread.lib;mkl_blacs_intelmpi_lp64.lib;libiomp5md.lib (MKL of Parallel Studio xe 2016.0.110)
My computer has 32Go (RAM) and 2 processors (Intel(R) Xeon(R) CPU E5-2687W v2 @ 3.40GHz).
Please find in attachment my program and associated files.