Short answer: we're building one for molecules
In your case, you don't actually want molecules. You're mentioning crystals. In that case, I suggest the Crystallography Open Database or if you have access, try the commercial Cambridge Crystal Structure Database. The COD is planning a theoretical crystal database, but I don't think it's online yet. These are experimental x-ray structures, but should be good starting structures for solid-state simulations.
For molecules, my group is building one. It's the Pitt Quantum Repository There are a few related efforts that I'm aware of.
With the PQR, you can either search through the website (even on mobile). The database is keyed by InChI keys, so if you have an existing structure you can generate the InChIKey and then get the structure, e.g.:
The long answer is that some commercial software vendors consider such libraries as violations of the license. No joke, I've planned this for years to augment Avogadro, but the company behind a leading program told me that they would void my license to distribute output files in this fashion.
With the rise of community and open source quantum chemistry software, this is moot. Use programs that let you distribute output files. What remained was:
- Gathering the 3D coordinates (i.e., a low-energy conformer) of X thousand/million molecules
- Obtaining the computer time to optimize all of them
- Devoting a server and disk space to serving them
- Having a great website to help view, browse, and search them