Concurrent Conda Environment Creation Latency

Goal: Determine how concurrent conda environment creation requests affect solver latency when the shared package cache resides on network‑attached storage.

Constraints: Use a recent Anaconda release (e.g., 2024.02), keep the package cache on a single NFS mount, vary the number of parallel `conda create` commands from 1 to 8, and record only the wall‑clock time reported by the underlying conda process without modifying solver settings.

Uncertainty: The exact latency increase depends on NFS performance, lock granularity introduced in conda 4.10, and whether the package index is static during the test.

Does enabling the experimental parallel SAT solver reduce the observed latency under concurrency? How does limiting concurrent solves to the number of physical CPU cores influence latency growth? What effect does moving the package cache to a local SSD have on the latency trend?