donaurelio / parallel-templates

C99 Templates for parallel programming

Parallel Templates

It is a set of templates of parallel programming patterns in C (for High Performance Compiting). For example: stencil to programing differential equation solvers and cellular automatas; map to replicate an operation or function over a collection, bettwen other parallel programing patterns.

Each template consists of 3 files: template.c, parallel.yml and context.yml

parallel-templates
│
└───templates
        ├── stencil
        │   ├── template.c
        │   ├── parallel.yml
        │   ├── context.yml
        │   └── Makefile 
        └── map
            ├── template.c
            ├── parallel.yml
            ├── context.yml
            └── Makefile 

Files Descrition

• template.c

...It is a code wirtten C, but with some jinja2 template syntax to add more flexibility to the user using the same template for different problems related with the same parallel programming pattern. This template must have a parallel programming pattern implicit. Additionally it must also be easily reusable.

• parallel.yml

...It is a metadata file in YAML format which in addition to having information of the template. It contains details of how it should be parallelized. That is, what OpenMP and OpenACC compile directives should go in each function and what loop lexicographically. This file is related with #pragcc tool.

• context.yml

...It is a metadata file in YAML format which contains values for the variables defined in a given template. This file can be modified by the user to get the same template with different values required by the problem to be specified in code.

• Makefile

...Contains the neccesary rules to compile and run the code, sequential and parallel.

Tests

The parallel_templates module is automatic tested by Travis CI, but you can perform test individually from the root directory ad follows.

python3 -m unittest tests/loader.py
python3 -m unittest tests/templates.py

API

This tool has an API to interact with the templates available in the application, at the moment it only has the stencil parallel programming pattern template. However, you can contribute to the growth of these.

The Api was developed with Flask Restplus. To run a API install requirements and run the application from the root directory ad follows:

pip3 install -r ./api/requirements.txt
python3 api/app.py

The API with run at http://localhost:5000

Deployment

The API can run inside a Docker Container. To get it done, perform the following commands from the root directory.

docker build -t templates .
docker run --rm -d -v ${PWD}:/usr/src/app -p 5001:5000 templates python3 api/app.py

References

Structured Parallel Programming: Patterns for Efficient Computation

Designing a RESTful API using Flask-RESTful

REST API response format using HTTP status codes

Flask-RESTful

Designing a RESTful API with Python and Flask

AJAX with jQuery

Flask Examples Github