Features includes NVIDIA V100 Tensor Cores, Full C++17 Language, PCAST Directives, Full OpenACC 2.6, New OS including macOS Mojave & more.
Overview
PGI Community Edition 19.4 is now available for free download, providing scientists and engineers with tools for high-performance computing (HPC). The new version includes enhanced support for CUDA Fortran, OpenACC directives, and full C++ 17 language support.
What You'll Learn
1
How to utilize CUDA Fortran support for NVIDIA V100 Tensor Cores
2
Why OpenACC 2.6 is important for performance portability
3
When to use OpenACC True Deep Copy Directives in your applications
Key Questions Answered
What are the new features in PGI Community Edition 19.4?
PGI Community Edition 19.4 introduces several new features including CUDA Fortran support for NVIDIA V100 Tensor Cores, full OpenACC 2.6 compliance, OpenACC True Deep Copy Directives, OpenACC printf(), PCAST Directives, full C++ 17 language support, and new OS support including macOS Mojave.
How does PGI support GPU computing?
PGI Compilers & Tools provide an easy on-ramp to GPU computing through the use of OpenACC directives, enabling developers to leverage GPU capabilities for high-performance computing applications effectively.
Technologies & Tools
Programming Language
Cuda Fortran
Used for developing applications that leverage NVIDIA GPU capabilities.
Directive
Openacc
Provides directives for parallel programming to enhance performance on GPUs.
Programming Language
C++ 17
Full language support is provided for developing modern C++ applications.
Key Actionable Insights
1Leverage CUDA Fortran support to optimize your applications for NVIDIA V100 Tensor Cores, which can significantly enhance computational performance.This is particularly beneficial for applications that require intensive numerical computations, allowing developers to fully utilize the capabilities of modern GPU architectures.
2Adopt OpenACC directives to improve the performance portability of your applications across different HPC platforms.Using OpenACC allows for easier code maintenance and scalability, as it abstracts the underlying hardware differences and lets developers focus on algorithm development.
3Implement OpenACC True Deep Copy Directives to manage data movement efficiently between host and device memory.This can help reduce latency and improve the overall performance of applications that rely heavily on data transfers in GPU computing environments.