There is no fixed limit in the plugin, but because of the way that Lightroom works it will use a lot of memory if you get it to scan a very large catalog for all duplicates.

In this situation, it would need to perform a large number of searches and this would take Lightroom days to complete. Instead, it creates its own indices which are much faster to access, but unfortunately do use lots of memory.

To work around this, scan the catalog in batches. Pick a group of photos and search for duplicates of them. For particularly large catalogs you may need to search only within the groups at first, and then in the entire catalog later once you have reduced the overall number of images.

As a rough guide, it is tested on machines with 4 and 6Gb of memory against a catalog with roughly 15,000 images, and will perform a full scan in a few minutes. I know of places running it against catalogs with many tens of thousands of images.