I can confirm that in very large applications decapsulation performance seems to deteriorate, but not to the extent you encounter. For instance, in a 200MB model on a 2.4MHz processor with 4GB, the studio allocated ~1GB or memory, decapsulation takes up to 15sec (except of-course for the special case that this is your first action in the model, which means the added time required to build the repository in memory).
Look into your configuration - perhaps allocating more memory to the studio, and minimizing the number of (other) processes running concurrently on your cpu.
Also, it makes sense to test the same functionality on another machine.
Please let us know your findings.
Notwithstanding, we do think decapsulation should be improved, although there is no immediate plan to do tackle this issue.
For best results, use the Firefox browser..