3 ms·
As the paper clarifies, this is for the performance of non-atomic read/writes in sequential setting. The paper left the performance evaluation for atomic read/w
by gadmm 5y ago
As the paper clarifies, this is for the performance of non-atomic read/writes in sequential setting. The paper left the performance evaluation for atomic read/writes to future work. Is there any indication yet regarding the performance in a parallel setting compared to weaker guarantees?
- avsm 5y agoThat's right -- we first wanted to establish that existing OCaml code wouldn't be adversely impacted. There are various efforts ongoing to build interested (locked and lock-free) concurrent data structures, so that will inform the parallel performance results. Nothing published yet.