https://github.com/thedavidchu/merry_mem
The Robin Hood hash table.
https://github.com/thedavidchu/merry_mem
hashtable robinhood
Last synced: 8 months ago
JSON representation
The Robin Hood hash table.
- Host: GitHub
- URL: https://github.com/thedavidchu/merry_mem
- Owner: thedavidchu
- License: mit
- Created: 2023-09-30T02:32:02.000Z (about 2 years ago)
- Default Branch: main
- Last Pushed: 2023-12-01T21:36:41.000Z (almost 2 years ago)
- Last Synced: 2025-01-04T05:26:33.300Z (9 months ago)
- Topics: hashtable, robinhood
- Language: C++
- Homepage:
- Size: 534 KB
- Stars: 2
- Watchers: 2
- Forks: 0
- Open Issues: 4
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
[](https://github.com/cpp-linter/cpp-linter-action/actions/workflows/cpp-linter.yml)
# merry_mem
This project includes a sequential and parallel version of the Robin Hood hash
table.
## Project Structure
The project is structured as follows:
```
|--src/
| |--common/ : Common utilities (types and a logger)
| |--parallel/ : Parallel implementation (library and simple sanity
| | check executable)
| |--sequential/ : Sequential implementation (library and simple sanity
| | check executable)
| `--trace/ : Code for generating traces to test our implementations
`--test/
|--performance_test/ : Benchmark the sequential vs the parallel parallel
| implementations with different traces and different
| numbers of workers
|--trace_test/ : Test the trace generating library
`--unit_test/ : Unit tests for our sequential and parallel
implementations
```
## Build
We use the CMake build system (version 3.18.4) that is available on the UG
machines. After ensuring you have CMake installed, build the CMake project:
```bash
# In the project's root directory
mkdir -p build
cd build
cmake -S .. -B .
```
This builds the required `Makefile` that allows you to build the project.
```bash
# In the build directory
make
```
To see the individual targets that can be made, run:
```bash
# In the build directory
make help
```
In general, executables are named `*_exe`.
## Test
After building the project, there are three types of tests:
1. Performance tests
2. Trace tests
3. Unit tests
Performance tests measure the timing of the sequential and parallel
implementations when running various traces.
```bash
#In the build directory
./test/performance_test/performance_test_exe
```
Trace tests ensure the trace generator produces traces in the expected format.
```bash
# In the build directory
./test/trace_test/trace_test_exe
```
Unit tests ensure that the sequential and parallel implementations match the
results of the same trace being run on `std::unordered_map`. To run the unit
tests, run:
```bash
# In the build directory
./test/unit_test/unit_test_exe
```
## Performance Test Plotting
After building the performance test, plot the performance graphs with:
```bash
# In the project's root directory
which python3
python3 test/plot/plot.py
```
This will create JSON files with the timing information and automatically
plot the sequential versus the parallel implementations as the number of worker
threads varies. These files will all appear in the project's root directory.