3 ms·
mOS is not directly related to the work described in that paper, but some of the stuff described there; e.g., reducing lock contention in physical memory page m
by rolfriesen 6y ago
mOS is not directly related to the work described in that paper, but some of the stuff described there; e.g., reducing lock contention in physical memory page management, is something we do in mOS as well. The lightweight kernel in mOS manages the physical pages, while we use Linux functionality to manage the virtual space. (This helps with compatibility.)
We have published several papers in the ROSS series of workshops, discussing mOS: https://www.mcs.anl.gov/events/workshops/ross/2020 https://www.mcs.anl.gov/events/workshops/ross/2020 Our 2018 IPDS (https://ieeexplore.ieee.org/document/8425166 https://ieeexplore.ieee.org/document/8425166) paper has some performance numbers, and the book here https://www.springer.com/gp/book/9789811366239 https://www.springer.com/gp/book/9789811366239 has a lot of detailed info and lessons learned from mOS and other Supercomputing OS projects around the world.