rbartlensky / rpools

Pool implementations to save memory

rpools

Build commands:

• cd benchmarks && python3 generate_alloc_file.py && cd .. - to include allocation benchmarks (optional)
• mkdir build && cd build && cmake .. && make - to build the project
• make install - to install the libraries
• make test - runs the tests of the project (this assumes that the project is built)
• pip install -r ./requirements.txt will install all the dependencies of all python scripts

plot_elapsed_time.py

This is used to plot the different outputs of the executables from benchmarks/elapsed_time/.

Usage (make sure you build the project first):

• python3 plot_elapsed_time.py -h to see the help menu

Examples:

• python3 plot_elapsed_time.py -b ./build/benchmarks/elapsed_time/bench_random -n 100000

Explanation (can also be found in the help menu of the script):

• ./build/benchmarks/elapsed_time/bench_random - the executable that we want to execute
• 100000 - the upperbound of objects that will be created (in the case of bench_worst, the upperbound is multiplied by the size of the pool)

plot_memory_usage

This is used to plot the output of the command: valgrind --tool=massif --stacks=yes --time-unit=ms on all the executables from benchmarks/memory_usage/.

Usage (make sure you build the project first):

• python3 plot_memory_usage.py -h to see the help menu

inject_custom_new

This is a script which is used to run an executable with a custom new/delete implementation. The custom new/delete library implements these operators by using the linked_pool/NSGlobalLinkedPool class to (de)allocate small objects and malloc to (de)allocate large objects.

Usage (make sure libcustomnew.so is installed):

• inject_custom_new my_exec args1 args2 - to run your executable with the custom implementation that is also thread safe

benchmarks/generate_alloc_file.py

This is a script that will generate allocation benchmarks. By running this script, the CMakeLists.txt file will change and next time make is called, the benchmarks will be compiled.

Note: Make sure you run this script from the benchmarks folder.

time_alloc_benchmarks.py

This is a script which is used to run all the allocation benchmarks.

Usage:

• python3 time_alloc_benchmarks.py -h for instructions.

Note: Make sure generate_alloc_file.py was run and the project was built.

Licenses!

Everything is GPLv3 except for the following files which have their own license:

• include/rpools/tools/light_lock.h (check source)
• src/avltree/avltree.c (check source)
• include/rpools/avltree/avltree.h (check source)
• src/allocators/MemoryPool.h (MIT)