[arielb1]

No. If the problem is computational resources, performance can't be increased by mere concurrency (while even completely deterministic STM-style parallelism would help).

Concurrency improves performance when a process accesses both the same computational resource and some other high-latency sharable resource.

[pron]

What is your "no" referring to? I totally agree with you. When I said "computational resource" I wasn't referring necessarily to CPU, let alone on the same machine, but resources of any kind. When one request is blocked, say, waiting for the DB (a computational resource), another can do some computation.