Yes, this is expected. One third-party library that we're using has a single slow method call that's not parallelized. That tends to dominate the runtime and is responsible for what you're seeing. You can expect to see some improvements to that picture (and in your runtime) in v1.2 from work that we've already done internally. However, that third-party library is going to keep things looking mostly like this for a while yet. Removing or improving it is a frequent topic of discussion 🙂