The Parallel Universe Magazine
Intel’s quarterly magazine helps you take your software development into the future with the latest tools, tips, and training to expand your expertise.
Issue 51, January 2023
Feature: The Case for SYCL*
In honor of the four-year anniversary of oneAPI, much of this issue focuses on SYCL* and oneAPI libraries like Intel® oneAPI Math Kernel Library (oneMKL) and Intel® oneAPI DPC++ (oneDPL).
Download Issue 51 to find:
- The feature article, in which Editor in Chief Henry Gabb explains why ISO C++ is not enough for heterogeneous computing.
- A look at how to modernize and accelerate C++ code using open, nonproprietary libraries
- An overview of how researchers ported to SYCL a critical scientific application used to model processes around black holes.
- Three articles that introduce AI applications powered by oneAPI, including a look at how the AI reference kit for Visual Quality Inspection is used in the pharmaceutical industry.
Contents
ISO C++ is not enough for heterogeneous computing. Learn how the lack of key features in C++ that are necessary to exploit accelerator devices stands in the way of using the language to create heterogeneous applications.
SYCL* has advantages for scientific computing. Learn how DPEcho, an MPI and SYCL version of a general relativistic magneto hydrodynamics application, can be used to improve the performance of modeling complex problems.
The Intel recommender system performs well in this challenge. Learn about Intel’s entry to this recommender system challenge: a graph neural network solution that won fourth place.
Get fast deep learning on your personal computer. Learn how to set up and use Stable Diffusion to convert text descriptions to images on your personal computer.
Use oneAPI to modernize and accelerate C++ applications while freeing them from vendor lock-in. Learn about the benefits of migrating Thrust applications to nonproprietary alternatives, and compare two migration strategies.
Use standard, open approaches to accelerate common math operations. Learn how to efficiently manage host-device data transfer while offloading math library functions to an accelerator using the OpenMP* heterogeneous parallel directives.
Learn about how the pharmaceutical industry uses an end-to-end computer vision and deep learning pipeline to identify defective pills in the manufacturing process.