Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.
Sign uponly advance the read cursor by one node when returning nodes #27
Conversation
This comment has been minimized.
Show comment
Hide comment
This comment has been minimized.
Show comment
Hide comment
This comment has been minimized.
Show comment
Hide comment
This comment has been minimized.
Show comment
Hide comment
ssiloti
Aug 24, 2016
Contributor
Doesn't seem worth it to me. Prefetch should hide most of the latency here and the transfer rate is nowhere near the memory bandwidth of a modern server. The node buffer shouldn't be getting evicted to disk during normal operation either.
|
Doesn't seem worth it to me. Prefetch should hide most of the latency here and the transfer rate is nowhere near the memory bandwidth of a modern server. The node buffer shouldn't be getting evicted to disk during normal operation either. |
ssiloti
closed this
Feb 21, 2017
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
arvidn commentedAug 24, 2016
to maximize the memory cache utilization. This is especially useful when serving nodes from a node buffer that gets evicted to disk.
Does this make sense?