I looking to run separate tasks in parallel that are required to share access to a few variables that are constantly updated in a program that is latency-sensitive. I am starting to see that this may not be currently possible in Julia.
The two approaches I have looked at so far appear to not achieve this.
Multithreading appears to be to focus on data parallelism or performing the same task on the same data in multiple threads via loop parallelism.
Parallel processes while able to perform to separate tasks appear to not have the ability to share access to data structures across processes without fetch. This is not truly sharing the data under my current understanding.
I am wondering if there is a third method or a method the above methods could be utilized to work in the method described?
It’s possible in version 1.3, see the recent juliacon talk on the subject
@Gavinclarke I do not know enough about your problem to make a meaningful comment.
However SharedArrays comes to mind
ps. I See you worked on a Kaggle challenge to identify salt domes from seismic data.
If you are interested I can put you in touch with people at CGG
Interesting, will have to give it a try. Thank you for the link.
I briefly looked into shared arrays, unfortunately, I really need other structures as the complexity of doing everything in just arrays would likely greatly increase the complexity.
Thank you for the offer about CGG. That Kaggle Challenge is really my only experience working with seismic data but I am interested in the field. However, due to this, I’m not sure of what use I would be currently. Thank you for your reply as well.