~/talk

Reconsidering OS Memory Optimizations in the Presence of Disaggregated Memory

Talkat ISMM 2022 · on Tuesday, 2022-06-14
links:none on file
links:none on file
links:none on file
slides:none on file

Claim: disaggregated memory with 750 ns–1.5 µs remote latency flips usual OS choices, so the authors build HotBox, a base-page-only memory manager that uses a hybrid access model (remote cache line accesses plus on-demand page migration) and an adaptive throttling scanner driven by LLC miss counters. Evaluation on an Intel PMAP emulator shows huge pages cause “hotness fragmentation” and can slow workloads, while HotBox outperforms Nimble and achieves near-native application time when local RAM holds ~40% of the working set at 750 ns.

AI-written summary · about

  1. 0:00Start
  2. 0:19Reconsidering Operating System Memory Memory Optimizations in the Presence of Disaggregated Memory
  3. 4:07Hybrid Access Model
  4. 6:28Impact of Huge Pages in Local Memory in the Disaggregated Memory
  5. 10:11Hotbox
  6. 11:08Adaptive Scanning and Migration Utility Mechanisms
  7. 22:22Live Questions
~/search
type to search 6,249 conferences, 138,987 speakers and 185,172 talks
~/help
cmdctrl k
search every conference, speaker and series
/
focus the filter search
h
home
c
conferences
s
speakers
e
series
t
light / dark — the status bar button does the same
?
this help
esc
close this, then clear every filter — or just the search, while you are typing in it

Ignored while you are typing. Pages without a filter rail have no search to focus.