Something might be working for me, because I don't see any files.
Could you describe in more detail your question? Yes it is possible to give LinearSolve a problem that is too large to be held in memory. You can use Sparse matrices when applicable to avoid this. Additionally, use Floating point numbers instead of rational/symbolic values to avoid using too much memory.
Are you able to run smaller examples?
