NCT-100: Programming Multicore Processors

• In-depth theoretical background covering multicore processor architecture, concurrent programming, parallel programming concepts and considerations.

• Critical concepts such as implicit and explicit parallelism, atomicity, synchronization, shared memory, cache coherency, Ahmdal's Law, Flynn's processor classifications and Little's law are covered in-depth.

• Recognize the best parallelism opportunities and explain the advantages of using threads to obtain concurrency using various analysis techniques, compositional approaches and parallel design patterns.

• How to avoid synchronization pitfalls such as starvation, deadlock, live lock and data races.

• How to implement and tune parallel algorithms.

• Define and use different synchronization methods effectively including mutexes/critical sections and conditional variables.

• Explain operating system interactions and the relationship between shared memory and threads.

• Explain what aspects of the operating system affect programming, how to deal with shared memory effectively, CPU selection, CPU-specific binding of threads, thread specific data and kernel-level scheduling.

• Understand and use threads with specific technologies and programming methods such as the Windows API, POSIX Pthreads, Intel TBB and OpenMP using C/C++ and the Intel Compiler

• Hands on experience with the Intel Compiler to build and run multithreaded programs during the hands-on learning laboratories and case studies.

• Best practices to deal with MT-unsafe libraries and how to write new thread-safe libraries.