The big disadvantage of locks is that performance decreases with contention, and performance of a shared-memory system in general degrades as the number of nodes increases. So every supercomputer in recent history uses a hierarchical approach: a network of multi-core units. Shared memory and locks for sharing data with cores in the same unit, message passing for sharing data with other units.
Just imagine trying to use system-wide locks on the IBM Sequoia. It has something like a million cores.
Also, advances have been made in manycore shared memory systems. The Cray XE6 (the hardware behind, e.g. HECToR [1]) has a hardware accelerated global address space with remote direct memory access that allows PGAS [2] to outperform MPI [3].
By the way, system wide locks are a red herring. At these scales, you avoid global data as much as you can, regardless of what your programming model is.
[1] http://www.hector.ac.uk/ [2] http://en.wikipedia.org/wiki/Partitioned_global_address_spac... [3] http://upc.lbl.gov/publications/pmbs11.pdf
The comment about global locks was intended to be silly, because comparing locks to message passing without talking about what you're doing with them is also silly.
The linked paper compares MPI message passing to an alternative hardware-accelerated message passing, which is interesting, but the choice of micro-benchmarks is not very exciting. To be clear, while the GP was really comparing the actor model (private memory + message passing) against the shared memory + locks model, I was only responding to the parent comment, and when I think "message passing" I don't automatically think "private memory".
You can have, say, a 100,000 x 100,000 matrix represented as an array over thousands of processors, where each processor can read and write each array element individually.