The 41st IPP Symposium

Towards a Common Language and Runtime Interface for Transactional Memory in C++

Ali-Reza Adl-Tabatabai, Intel

The transactional memory community is now at a stage where it needs developers to experiment with writing large programs using transactional memory programming constructs. This is important not only to prove and refine the TM programming model, but also to tune the performance and scalability of TM implementations on non-trivial workloads. To facilitate writing large applications using TM, several groups have released compilers that support language extensions for TM in C or C++. In this talk, I argue that incompatible language interfaces will hamper adoption of TM. To foster adoption of TM and bootstrap development of large applications, I propose a common, core set of C++ language extensions for TM, enabling developers to write applications that will work with different TM compilers. I further propose a common compiler-runtime ABI that would enable binaries generated by different TM compilers to interoperate and would allow interoperation between different compilers and TM runtimes.

Bio: Ali-Reza Adl-Tabatabai is a Senior Principal Engineer in Intel's Programming Systems Lab. He leads a team of researchers working on compilers and scalable runtimes for future Intel Architectures. His current research concentrates on language features supporting parallel programming for future multi-core architectures and on architectural support for those features. Most recently he has worked on transactional memory. Ali received his Ph.D. in Computer Science from Carnegie Mellon University.