Codeplay SYCL 1.2.1 Solution offers an Open Alternative to CUDA

Print Friendly, PDF & Email

Today Codeplay announced the world’s first fully-conformant SYCL 1.2.1 Solution.

As a non-proprietary alternative to the incumbent CUDA, SYCL is an open standard developed by the Khronos Group that enables developers to write code for heterogeneous systems using standard C++. Developers are looking at how they can accelerate their applications without having to write optimized processor specific code. SYCL is the industry standard for C++ acceleration, giving developers a platform to write high-performance code in standard C++, unlocking the performance of accelerators and specialized processors from companies such as AMD, Intel, Renesas and Arm.

Codeplay’s ComputeCpp 1.0 enables SYCL and provides C++ developers with huge benefits:

  • High Performance Computing: Supercomputers are playing an important role in computationally intensive tasks in the fields of science, finance, and many others to provide complex calculations and simulations. SYCL offers a standard way for HPC developers to write portable, efficient, accelerated code using standard C++ that can be deployed to GPUs, FPGAs and other accelerators
  • Computer Vision: Complex image processing operations can be accelerated using parallel computing. ComputeCpp and SYCL provide high-level programmability for custom vision processors, enabling additional custom features on top of existing optimized hardware functions
  • Artificial Intelligence: Linear algebra is increasingly being used in artificial intelligence applications and benefits from parallel architectures. The Eigen linear algebra librarySYCLBLAS and TensorFlow frameworks can be accelerated using ComputeCpp for a wide variety of heterogeneous hardware

Codeplay continues to lead the SYCL community, as demonstrated by the release of ComputeCpp 1.0 featuring the industry’s first fully conformant SYCL 1.2.1 implementation,” said Codeplay’s Michael Wong, chair of the SYCL working group and senior member of the ISO C++ Standards committee. “This latest release will empower developers to more easily harness the power of heterogeneous computing. It also helps improve the experience for developers by making it easy to write applications that can take greater advantage of the compute capabilities present in a range of CPUs, GPUs, and other processor types.”

Last year, Renesas announced its commitment to enabling SYCL with ComputeCpp for the R-Car family of hardware. This enables developers of Advanced Driver Assistance Systems (ADAS) to build complex computer vision and artificial intelligence software with minimal porting and development effort, accelerating their time to market. This will be instrumental in bringing artificial intelligence to the next generation of cars, including autonomous vehicles.

The recent emergence of machine learning and deep neural networks has uncovered a new use for GPUs in particular, and SYCL has proved to be an enabler to accelerating the execution of many machine learning applications. Deep learning frameworks, including TensorFlow, can use the SYCL interface to offer significant performance improvements by using heterogeneous systems.

ComputeCpp will enable advanced vision and machine learning developers to use the familiar C++ programming environment and the SYCL standard to quickly benefit from a range of heterogeneous hardware including R-Car’s advanced IMP-X5 computer vision IP.” said Andrew Richards, CEO at Codeplay. “We are experiencing a substantial increase in momentum and enthusiasm for the many benefits offered by SYCL.”

ComputeCpp 1.0 Key Features:

  • Conformant SYCL 1.2.1 Implementation Develop with familiar tools and standard C++
  • Cross-platform, single-source, high-level C++ programming layer
  • Built on top of OpenCL and based on C++14
  • Delivers a heterogeneous programming solution for C++
  • Write portable code that runs on multiple hardware platforms
  • C++ Support Use templates for generic programming
  • Lambdas allow for seamless single-source development
  • RAII for better resource management
  • Access to low-level memory management for better memory management optimizations

The developer community continues to get behind SYCL as the best way to optimize applications and leverage the unmatched level of compute processing capabilities of processor acceleration.

Developers can download ComputeCpp 1.0 from the Codeplay developer website and see the latest news and videos on SYCL at http://sycl.tech.

In related news, Codeplay today launched of a set of open source learning materials for SYCL, called the “SYCL Academy.”

Sign up for our insideHPC Newsletter