Simplify Coding Across Architectures with DPC++ | Intel Software

Опубликовано 5 февраля 2021, 0:33

See how Data Parallel C++ (DPC++) is an open, standards-based, cross-architecture language that is part of oneAPI. It simplifies heterogeneous programming across multiple types of architectures (CPUs, GPUs, FPGAs, and specialized accelerators) without proprietary lock-in, providing developers freedom of choice in hardware. DPC++, an open-community project, is based on C++ and contains SYCL to support parallelism and heterogeneous programming. The SYCL kernel-based programming model helps application developers target a variety of accelerated computing platforms. New extensions created with the community bring additional capabilities and increase efficiency.

DPC++ interoperates with other languages, including C, C++, Fortran, and Python, as well as parallel models, such as OpenMP and MPI. It’s designed to realize the full value of hardware, delivering native high-level language performance - unlike managed runtime cross-architecture solutions. DPC++ also provides outstanding programming productivity and allows code reuse across hardware targets while permitting custom tuning for a specific accelerator. This frees developers to focus on optimization and innovation - not rewriting software for the next hardware platform.

Use oneAPI, DPC++, and Intel’s oneAPI Toolkits, in the Intel® DevCloud where developers can develop, test, and run workloads on a cluster of the latest Intel hardware: sforce.co/2XRX9AU

Subscribe to the Intel Software YouTube Channel: bit.ly/2iZTCsz

Learn more about oneAPI: oneapi.com
DPC++ Developer Resources
Overview: software.intel.com/content/www...

Training: software.intel.com/content/www...

Documentation: software.intel.com/content/www...

oneAPI Toolkits: software.intel.com/content/www...

Get more info on software development tools: twitter.com/inteldevtools

Simplify Coding Across Architectures with DPC++ | Intel Software
youtube.com/intelsoftware