That’s more than fair.

In the system I’m most familiar with, however, the benefits of hyperthreading for throughput extend to the 50-70% utilization band where p99 latency is not stressed.