The current generation of processors has abandoned ever-increasing frequencies as the path to unlimited performance gains and throws programmers into the bewildering world of parallel computation. Multithreaded applications present a host of subtle pitfalls to developers unfamiliar with their unique difficulties. Even worse, applications which don't step into this brave new world will fail to benefit from the enhanced capabilities of their execution platform. Fortunately, the GNU Compiler Collection offers tools to safely parallelize existing code and promises even more exciting automated optimizations in the future.

Automatic multithreading tools allow programmers to reap the benefits of multicore architectures without changing a single line of source code. Future releases of GCC promise even greater performance gains through more intelligent optimizations.

Perhaps even more important, compiler optimizations shift some of the burden of concurrent design from the application developers to compiler writers. Powerful compile-time analysis guarantees programmers can sidestep non-deterministic bugs like deadlock and starvation in many projects.

This talk will examine the current and future state of automated multithreading in GCC, with a focus on its implementation on the Linux platform.