If OpenBLAS is built using add_subdirectory(OpenBlas) as part of another project
then the paths set by CMAKE_XXX_DIR are relative to the parent project
and not the OpenBLAS project.

The NTRA argument can be equal to -1 if one does not want a snapshot file
(and this is the case with sample data {s,d,c,z}in3).
The routines {S,D,C,Z}PRCN3 will try to use their first argument as an output
unit number, so we avoid calling them when NTRA < 0.

Patch originally written by Camm Maguire.

The SNAME variable contains names of C functions like "cblas_dgemv".
Apparently the code was not taking into account the 6-letter "cblas_"
prefix when determining the task to be done.

The issue does not affect c_{s,d,c}blat2.f, which use the correct
offsetting.

Patch originally written by Camm Maguire.

Signed-off-by: Timothy Gu <timothygu99@gmail.com>

When OpenBLAS is compiled with USE_OPENMP=1, it ignores OPENBLAS_NUM_THREADS and GOTO_NUM_THREADS flags.Therefore, you should use OMP_NUM_THREADS.

Without setting OMP_NUM_THREADS, a process will use maximal number of threads on a computing node. Thus, if there are 2 processes on the computing node, the thread will contend against other threads on CPU cores. As a result, the application will hang.