A preliminary assessment of Haskell's software transactional memory constructs

In recent years, usage of multicore architectures has boosted the need for concurrency. Many researchers have pointed out Software Transactional Memory (STM) as an easy way to achieve it. This study presents a preliminary evaluation of Haskell's STM constructs against its lock-based alternative. Subjects which used STM committed significantly fewer mistakes and required on average 12% less time to finish their assignments.