Today, Intel announced the 2023 release of the Intel oneAPI tools - available in the Intel Developer Cloud and rolling out through regular distribution channels. The new oneAPI 2023 tools support the upcoming 4th Gen Intel Xeon Scalable processors, Intel Xeon CPU Max Series and Intel Data Center GPUs, including Flex Series and the new Max Series. The tools deliver performance and productivity enhancements, and also add support for new Codeplay plug-ins that make it easier than ever for developers to write SYCL code for non-Intel GPU architectures. These standards-based tools deliver choice in hardware and ease in developing high-performance applications that run on multiarchitecture systems.

"We're seeing encouraging early application performance results on our development systems using Intel Max Series GPU accelerators - applications built with Intel's oneAPI compilers and libraries. For leadership-class computational science, we value the benefits of code portability from multivendor, multiarchitecture programming standards such as SYCL and Python AI frameworks such as PyTorch, accelerated by Intel libraries. We look forward to the first exascale scientific discoveries from these technologies on the Aurora system next year."
-Timothy Williams, deputy director, Argonne Computational Science DivisionWhat oneAPI Tools Deliver: Intel's 2023 developer tools include a comprehensive set of the latest compilers and libraries, analysis and porting tools, and optimized artificial intelligence (AI) and machine learning frameworks to build high-performance, multiarchitecture applications for CPUs, GPUs and FPGAs, powered by oneAPI. The tools enable developers to quickly meet performance objectives and save time by using a single codebase, allowing more time for innovation.

This new oneAPI tools release helps developers take advantage of the advanced capabilities of Intel hardware:

• 4th Gen Intel Xeon Scalable and Xeon CPU Max Series processors with Intel Advanced Matrix Extensions (Intel AMX), Intel Quick Assist Technology (Intel QAT), Intel AVX-512, bfloat16 and more.
• Intel Data Center GPUs, including Flex Series with hardware-based AV1 encoder, and Max Series GPUs with data type flexibility, Intel Xe Matrix Extensions (Intel XMX), vector engine, Intel Xe Link and other features.

Example benchmarks:

• MLPerf DeepCAM deep learning inference and training performance with Xeon Max CPU showed a 3.6x performance gain over Nvidia at 2.4 and AMD as the baseline 1.0 using Intel AMX enabled by the Intel oneAPI Deep Neural Network Library (oneDNN).
• LAMMPS (large-scale atomic/molecular massively parallel simulator) workloads running on Xeon Max CPU with kernels offloaded to six Max Series GPUs and optimized by oneAPI tools resulted in an up to 16x performance gain over 3rd Gen Intel Xeon or AMD Milan alone.

Advanced software performance:

• Intel Fortran Compiler provides full Fortran language standards support up through Fortran 2018 and expands OpenMP GPU offload support, speeding development of standards-compliant applications.
• Intel oneAPI Math Kernel Library (oneMKL) with extended OpenMP offload capability improves portability.
• Intel oneAPI Deep Neural Network Library (oneDNN) enables 4th Gen Intel Xeon and Max Series CPU processors' advanced deep learning features including Intel AMX, Intel AVX-512, VNNI and bfloat16.

To boost developer productivity, enriched SYCL support and robust code migration and analysis tools make it easier to develop code for multiarchitecture systems.

• The Intel oneAPI DPC++/C++ Compiler adds support for new plug-ins from Codeplay Software for Nvidia and AMD GPUs to simplify writing SYCL code and extend code portability across these processor architectures. This provides a unified build environment with integrated tools for cross-platform productivity. As part of this solution, Intel and Codeplay will offer commercial priority support starting with the oneAPI plug-in for Nvidia GPUs.
• CUDA-to-SYCL code migration is now easier with more than 100 CUDA APIs added to the Intel DPC++ Compatibility Tool, which is based on open source SYCLomatic.
• Users can identify MPI imbalances at scale with the Intel VTune Profiler.
• Intel Advisor adds automated roofline analysis for Intel Data Center GPU Max Series to identify and prioritize memory, cache or compute bottlenecks and causes, with actionable insights for optimizing data-transfer reuse costs of CPU-to-GPU offloading.

Why It Matters: With 48% of developers targeting heterogeneous systems that use more than one kind of processor, more efficient multiarchitecture programming is required to address the increasing scope and scale of real-world workloads. Using oneAPI's open, unified programming model with Intel's standards-based multiarchitecture tools provides freedom of choice in hardware, performance, productivity and code portability for CPUs and accelerators. Code written for proprietary programming models, like CUDA, lacks portability to other hardware, creating a siloed development practice that locks organizations into a closed ecosystem.

About oneAPI Ecosystem Adoption: Continued ecosystem adoption of oneAPI is ongoing with new Centers of Excellence being established. One, the Open Zettascale Lab at the University of Cambridge, is focused on porting significant exascale candidate codes to oneAPI, including CASTEP, FEniCS and AREPO. The center offers courses and workshops with experts teaching oneAPI methodologies and tools for compiling and porting code and optimizing performance. In total, 30 oneAPI Centers of Excellence have been established.