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 49, August 2022

Feature: SYCLomatic—A New CUDA* to SYCL* Code Migration Tool

This issue focuses on optimizing parallel programming and end-to-end pipelines while eliminating vendor lock-in using the oneAPI ecosystem, SYCL*, and other tools.

Download the current issue to find:

  • The feature article describing a new open source project that makes it easier to migrate CUDA* GPU code to SYCL.
  • Intel acquired Codeplay Software*. Editor in chief and senior principal engineer Henry Gabb discusses what it means for SYCL and the oneAPI ecosystem.
  • Real-world life sciences applications and the tools and methods that improve power efficiency and performance while reducing cost.
  • A demonstration of best practices for building oneAPI for Linux* and Windows* using the Windows Subsystem for Linux 2 and Microsoft Visual Studio* Code.
Contents
SYCLomatic: A New CUDA to SYCL Code Migration Tool

Discover the SYCLomatic open source project, a tool that makes it easier to migrate CUDA GPU code to SYCL. Submit your contributions to this project.

The SigOpt* Intelligent Experimentation Platform

Explore end-to-end democratization solutions that improve pipeline efficiency of recommender systems.

Deliver Cost-Effective Genomics for Precision Medicine

Compare Sentieon* DNASeq performance to NVIDIA Clara Parabricks*. Explore how Sentieon reduces the cost and time required to analyze whole genomes and exomes.

Leverage Tools for Cross-Platform Software Development

Learn from technology evangelist Tony Mongkolsmai how modern software developers can be more productive and spend less time fighting with their tools. He also shows how to build oneAPI applications for Linux and Windows.

Free Your Software from Vendor Lock-in Using SYCL and oneAPI

Migrating from CUDA* to SYCL is now easier. Learn how to use SYCL and oneAPI to write parallel software that can make use of GPUs and other accelerators without vendor lock-in.

Distributed Training on Intel® Xeon® Scalable Processors

This case study presents training an AI model on an AI Arena platform from Tencent*. Find out how to efficiently train complex deep learning models using distributed training with TensorFlow* and Horovod*.

Accelerate Single-Cell Genetics Analysis

End-to-end Scanpy genetics pipelines are nearly 1.5x faster with Intel® Xeon® processors than NVIDIA* A100, significantly reducing computing costs.