Lockless Circular Buffer over Shared Memory (High Speed Data Transfer via Shared Memory)

With Real-time and Embedded Software development moving from single core to multi-core platforms, the era is now dominated with multi-threaded software. New class of Software which attempts to leverage advanced multi-core targets have multiple streams of execution (applications/threads) running independently of others. With applications running on different cores, the requirement is for a robust mechanism to exchange data between the applications at high speeds with minimal CPU overhead and minimal performance penalty. Traditionally, one of the Solutions has been Shared Memory or SHM with the drawback of using Semaphores to control access of shared regions. But this makes the system blocking and effects performance. This paper proposes an improved High Speed Data Transfer mechanism over SHM while keeping up the system lockless.