It is recommended to ensure that your nodes have a sufficient amount of CPU cores at their disposal. The project documentation[1] states:
The Cassandra write path tends to be heavily optimized (writing to the commit log and then inserting the data into the memtable), so writes, in particular, tend to be CPU bound. Consequently, adding additional CPU cores often increases throughput of both reads and writes.
Between 8 and 12 CPU (24 hyper-threaded) cores should be sufficient for a production deployment.