This READ.md template was written based on this repository.
- Murillo Boratto, our mentor, your experience and knowledge have been invaluable to our progress.
This repository contains code samples in DPC++, an extension of the C++ standard created by Intel® for heterogeneous parallel programming. The code samples were created with the aim of helping developers understand and use the DPC++ language in their projects.
DPC++ is an extension of the C++ standard that enables heterogeneous parallel programming on systems with multiple types of processors, such as CPUs and GPUs. It was created by Intel® in collaboration with other companies and is based on open-source technologies like LLVM and SYCL.
- Collection of basic parallel DPC++ code examples.
- Instructions and information targeted at beginners in parallel programming.
- Concepts explanations like array reduction, matrix multiplication, and vector addition using DPC++.
To use the code samples, you need to have a compiler that supports DPC++, such as the Intel® DPC++ Compiler or hipSYCL. You also need a system with support for heterogeneous processors, such as an Intel® CPU and an AMD Radeon GPU. The project dependencies are described in a dependencies file within the repository. Dont forget to download DPCPP and Sycl libraries within the Intel® oneAPI DPC++/C++ Compiler to run the codes inside this repository.
To compile and run the examples in this repository, inside the codes directory, execute the following commands, where output-name refers to the name the executable file will have after compilation, and source-code refers to the name of the code file you want to compile:
-
Load the oneAPI environment with the following command:
source /opt/Intel®/oneapi/setvars.sh -
To compile, use the following command:
dpcpp -o output-name source-code -
Running the code:
./output-name -
[ Example ] To compile the arrayReduction, we need to run the following commands:
source /opt/Intel®/oneapi/setvars.sh dpcpp -o arrayReduction ./arrayReduction.cpp ./arrayReduction
Developers worked for two weeks using the Kanban methodology to develop this project. Each week, they synchronized and delivered their tasks.
Feel free to create a new branch, fork the project, create a new Issue or make a pull request contact one of us to develop at DPCPP for dummies.
[1] INTEL. Guia do Desenvolvedor e Referência do Compilador Intel® oneAPI DPC++/C++. [S.l.], 2023. Available at: https://www.intel.com/content/www/us/en/docs/dpcpp-cpp-compiler/developer-guide-reference/2023-0/overview.html. Accessed on: June 16, 2023.