Application-agnostic NUMA tuning, including allocator replacement, memory interleaving, and disabling AutoNUMA and THP, substantially speeds up in-memory analytics workloads.
We explore application-agnostic 3 strategies that can be applied to the data analytics applica- tion in either a black box manner, or with minimal tweaks to the code
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DB 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Toward Efficient In-memory Data Analytics on NUMA Systems
Application-agnostic NUMA tuning, including allocator replacement, memory interleaving, and disabling AutoNUMA and THP, substantially speeds up in-memory analytics workloads.