As @billyc pointed out, .csv files are inefficient for large networks. We have started using SpatialLite for working with large networks, as it is more efficient, and can store several GMNS objects into one database. To accommodate efficient storage options, recommend that GMNS only specify the type of geometry (e.g., points, lines), while not dictating how the geometry is stored (e.g., x/y coordinate, WKT, etc.)
As @billyc pointed out, .csv files are inefficient for large networks. We have started using SpatialLite for working with large networks, as it is more efficient, and can store several GMNS objects into one database. To accommodate efficient storage options, recommend that GMNS only specify the type of geometry (e.g., points, lines), while not dictating how the geometry is stored (e.g., x/y coordinate, WKT, etc.)