Within each enclave is a thread, the basic runtime path represented by the machine state; conditions raised during execution are isolated to that runtime path.

Threads share all of the resources of an enclave and therefore do not need to selectively create or load new copies of resources, code, or data. Although a thread does not own its storage, it can address all storage within the enclave. All threads are independent of one another and are not related hierarchically. A thread is dispatched with its own runtime stack, instruction counter, registers, and condition handling mechanisms.

Because threads operate with unique runtime stacks, they can run concurrently within an enclave and allocate and free their own storage. Concurrent, or parallel, processing, is useful when code is event-driven, or for improving the performance of a large application.