Thanks again. If I am right though, the memory barriers could also have been skipped without thread-local values since, due to the barriers on init, no thread's view would revert to it's pre-init state anyway. As rbehrends mentions, barriers are not the only factor... For example, global memory access may be slower on some systems due to decreased locality (a property provided by the TLS), but I don't know what if any value that would have in this benchmark. There is also a comment on that page that does the same thing as my TLS-free approach using a more reasonable static global initialization syntax in Java in response to a question of if D's initialization would be semantically equivalent too.
Edit: (Not waiting for reply link to appear...) When I say "the memory barriers could also be skipped" I am referring to any other memory barriers past those associated with initialization, not those associated with initialization. There would be memory barriers at initialization, but that is the only need for memory barriers and I believe that is provided here by the synchronized keyword. The object presented in the article then looks like a cache-friendlier version of a single global variable with the necessary barriers limited to the initialization of the global instance pointer (which might also be done as in the Java example given in the comments on the page). The TLS does not eliminate the need for any barriers, only helps eliminate non-local reads. That might allow a performance gain depending on the cache behavior (or, for example, in the presence of NUMA), but it hasn't reduced the need for locking IMHO.