xiaorz / pibench-ep2

Benchmarking post-Optane persistent memory range indexes

Benchmarking New Persistent Indexes

Here hosts resources for benchmarking newly proposed persistent indexes on Pibench, including benchmark scripts (benchmark.sh & benchmark.py), source code of 6 indexes (see below) and pre-compiled pibench wrappers (wrappers/).

See detailed analysis in our tech report here: https://arxiv.org/abs/2201.13047

Dependencies

Pibench [https://github.com/sfu-dis/pibench.git] is needed to run any benchmark.
PMDK [https://pmem.io/pmdk/] is required to run most index under persistent mode.

Python benchmark scripts

cp -r wrappers /youPathTo_PiBench_Build_Src 
cp benchmark.py /youPathTo_PiBench_Build_Src
python3 benchmark.py

Installation

Pre-compiled shared libs can be found under wrappers directory.
The README file in each index folder also contains instructions for configuring and producing .so file

1. DPTree
2. FPTree
3. HOT
4. LB+-Tree
5. Masstree
6. ROART
7. utree