Realtime-NMR-Multicore-MixedCriticality-Scheduler

Sharif University of Technology - Real Time Systems project

Team

• Saee Saadat
• Neda Taghizadeh

Simulator

Usage

python run.py <config_number> <mode>

examples:

python run.py # To run everything
python run.py configs/1.1.yml
python run.py configs/1.1.yml plot
python run.py configs/1.1.yml edf-only

---

Stats:

Visit final_records.yml and final_records_full.yml for the final results of the simulations. You can find relevant logs in the logs folder. Please visit the charts folder for some generated charts.

---

Task Generator

Usage

python generate.py <config_number> <mode>

the mode argument is optional and can be either debug or report. If the mode is debug, the program will print the results to the console. If the mode is report, the program will generate two csv reports in the reports folder. One for the tasks and one for the core assignments. the name of the report will correspond to the config file and the assignment method used for the tasks.

Config

The config file is a YML file that contains the following fields:

• random_seed: The seed used for generating random numbers. (Optional for consistent results)
• num_of_tasks: Number of tasks to be generated
• ratio: Ratio of the number of high criticality tasks to the number of low criticality tasks
• num_of_cores: Number of cores in the system
• core_utilization: Utilization of each core
• periods: A list of periods for the tasks (chosen uniformly at random)
• mu_range: the minimum and maximum values for the mu parameter in the UUniFast algorithm
• overrun_percentage: The percentage of cores that are in overrun state
• error_rate: The error rate of the system (what percentage of cores might fail their tasks)
• reliability: The required reliability of the system
• assignment_policy: The policy used for assigning tasks to cores. (WFD or FFD)
• scheduling_policy: The policy used for scheduling the tasks. (Only EDF-VD has been implemented)

UUniFast task generation

Task generation based on the UUniFast algorithm is done based on This Paper Since we are using NMR, the utilization generated for HC tasks is divided by the number of copies needed to achieve the required reliability.

Calculating the number of copies needed to ensure reliability

using this formula:

N = ceil(-log(1 - R) / log(P)) if P ≠ 0, else 0

The number of copies we need for a task with failure rate of P to achieve reliability of R is calculated.

Task Assignment

The tasks are assigned to cores based on the chosen policy. The policies implemented are WFD and FFD.