https://github.com/openworm/multi-dev-sibernetic
Multi devices sibernetic engine
https://github.com/openworm/multi-dev-sibernetic
Last synced: about 1 year ago
JSON representation
Multi devices sibernetic engine
- Host: GitHub
- URL: https://github.com/openworm/multi-dev-sibernetic
- Owner: openworm
- License: mit
- Created: 2023-01-09T07:40:49.000Z (over 3 years ago)
- Default Branch: main
- Last Pushed: 2023-01-20T10:24:58.000Z (over 3 years ago)
- Last Synced: 2025-02-16T07:16:37.051Z (over 1 year ago)
- Language: C++
- Size: 3.8 MB
- Stars: 1
- Watchers: 12
- Forks: 0
- Open Issues: 2
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# multi-dev-sibernetic
This repo for multi devices sibernetic engine implementation. This implementation was testen on Novosibirsk State University cluster on configuration with several GPU unit from 1 to 8. GPU - NVIDIA Tesla V100. Results was reported [here](http://nusc.nsu.ru/wiki/lib/exe/fetch.php/reports/sskhayrulin/nsu_iis_sskhayrulin_202211.pdf) (in Russian), and [here](http://nusc.nsu.ru/wiki/lib/exe/fetch.php/reports/sskhayrulin/nsu_iis_sskhayrulin_202006.pdf) (in Russian) this old report.
# How to run
## Build
For building you have to run command `make -f makefile` it will create bin folder and create executable file
To clean all building arifacts you have to run command `make clean`
### Run options
```
Usage: ./bin/sibernetic [OPTION]
--multi_dev Run with all available devices.
but default it will run only one.
-f Path to configuration file.
-help, -h, -?, --help Print this information.
--p_sort turn on parallel sorting.
-iter_limit= limit for number of iteration simulation will preferme before stop.
-time_limit=0.5 limit of time simulation in seconds (will run 0.5 seconds of simulation).
Full documentation at:
Please report any bugs/issues to:
```
# Algorithm
Add it soon
...
# Publication
Main algorithm idea was published [here](https://nccbulletin.ru/files/article/article-_5_bul_45_29_07_0.pdf)