3 ms·
An Atomic Hash Table
- th0br0 9y agoThere's also https://github.com/facebook/folly/blob/master/folly/docs/AtomicHashMap.md https://github.com/facebook/folly/blob/master/folly/docs/Ato... that I know of trying to solve this problem.
- IIAOPSW 9y agoThis might be just what I need. I had this idea for a quantum computer simulation which stores states similar to how the physicist writes them. Eg a|00> + b|11> becomes {(0,0):a, (1,1):b}. In general sums of kets, bras, operators are represented by dictionaries. Implementing operations on a QC is a very parallelization task, but is hard to do concurrently with built-ins.
- Terretta 9y agoNote, this is a component of the sibling front page article on TFS file system here: https://news.ycombinator.com/item?id=14386331 https://news.ycombinator.com/item?id=14386331
- eternalban 9y agoI've always admired the wizardry of Cliff Click's concurrent hash: http://www.azulsystems.com/blog/cliff/2007-03-26-non-blocking-hashtable http://www.azulsystems.com/blog/cliff/2007-03-26-non-blockin... https://youtu.be/HJ-719EGIts https://youtu.be/HJ-719EGIts I'm not sure what to make of an "atomic" hash. I don't think OP does as well: "Doesn't this kind of defeat the point of concurrent algorithms? Well, in a sense, it does make it non-atomic, but it isn't really a spin-lock or anything like that, even though it can seem like that:" Concurrent hash seems to be more correct and clear.