We anticipate releasing an implementation of ICC++ in late 1995. Our current implementation of ICC++ is operational on uniprocessors, the TMC CM-5, and the Cray T3D machines with the performance described in this chapter. However, because the implementation has only been operational for a few months, we are still implementing a number of C++ specific optimizations which were not previously required in the Concert system. Future efforts will include novel optimizations for efficient parallel code generation, concerted optimization across the compiler and runtime, working and optimized versions of the ICC++ system for distributed shared memory machines and clustered symmetric multiprocessor machines, as well as extensive application studies. In fact the application studies are already well underway, and one large application and numerous kernels are being ported to ICC++ already.