1 paper
Sean Siddens, Sanya Srivastava, Reese Levine +2
To improve efficiency, nearly all parallel processing units (CPUs and GPUs) implement relaxed memory models in which memory operations may be re-ordered, i.e., executed out-of-orde…