Intel oneAPI 2023.2 now available
News|by Leanne Bevan|26 July 2023
Intel has unveiled the latest version of its oneAPI tools, version 2023.2, which promises to enhance software development for open multivendor multiarchitecture accelerated computing. This update comes with several standout enhancements, making it an exciting release for developers and hardware vendors alike.
What's new in Intel oneAPI 2023.2?
CodePlay
One of the notable updates is the addition of SYCL support for NVIDIA and AMD platforms through Codeplay's plug-ins. This integration brings the benefits of open, heterogeneous computing to a wider range of hardware, fostering better performance and flexibility in software development.
Codeplay has also made two other important announcements in conjunction with the 2023.2 release. Firstly, they contributed uniquely valuable technology from the computecpp SYCL compiler project to the open source DPC++ SYCL compiler project. This collaboration aims to bring SYCL into the LLVM world, encouraging more developers to embrace this technology. Secondly, Codeplay introduced the oneAPI Construction Kit, designed to assist hardware vendors in adopting multivendor multiarchitecture accelerated computing effectively.
Python, AI and HPC
One of the most eagerly awaited enhancements in the 2023.2 release is the faster Python support for AI and HPC applications. The accelerated NumPy functions, along with Numba compiler support, offer performance boosts for Python applications. Intel plans to expand on this feature even further in its upcoming 2024.0 release.
Additionally, AI inferencing becomes faster and more accurate with the support for "missing values" (NaN) during inference, streamlining pre-processing and improving prediction accuracy for models trained on incomplete data.
oneDNN library
The update also includes improvements to the oneDNN library, with optimisations and a new experimental Graph Compiler backend for the Graph API. These enhancements target specific processors with Intel AVX-512 and Intel AMX instructions, aiming to boost performance for AI-related tasks.
GPU support
Furthermore, Intel's Open Image Denoise now supports Intel, Nvidia, and AMD GPUs, providing more choices for developers working on AI-based image enhancement tasks across different hardware platforms.
This release also highlights impressive performance improvements for Intel GPUs, including ray-tracing support, showcasing the company's commitment to optimising tools and libraries for its hardware.
Performance enhancements
The 2023.2 release brings numerous performance enhancements, including faster SYCL compilation, improved math library performance, and standard C++ parallelism routines for the Intel Data Center GPU Max Series. Moreover, the SYCLomatic open source tool, designed to aid migration from CUDA to SYCL, has been improved with expanded support for CUDA APIs, FP64, and the latest version of CUDA.
The oneAPI Threading Building Blocks (oneTBB) algorithms and Flow Graph nodes now accept new types of user-provided callables, resulting in a more powerful and flexible programming environment through the use of std::invoke.
For Intel Fortran Compiler users, the extension of support for Do Concurrent Reduction brings a powerful feature that significantly improves the performance of code performing reductions while making parallel code writing more efficient and error-free by detecting uninitialised variables using the LLVM memory sanitiser.
The Intel Cryptography Primitives Library now supports the XTS mode of the SM4 algorithm, providing efficient and secure data encryption for storage devices.
Furthermore, the Intel Distribution for GDB has been rebased to GDB 13, adding support for Shared Local Memory (SLM) and staying current with the latest debugging enhancements.
Lastly, the release introduces "Velocity Bench for GPU Offload Performance Data," which aims to establish cross-platform benchmarks for GPU performance comparison. This effort aligns with the goal of oneAPI to make all hardware accessible to all applications, irrespective of whether they are open source or closed source.
Buy Intel oneAPI 2023.2 with Priority Support
Grey Matter is an Intel Elite Reseller. We can provide Intel oneAPI Priority Support licensing.
Please fill out the form below for a quote or more information from one of our team.
Contact Grey Matter
If you have any questions or want some extra information, complete the form below and one of the team will be in touch ASAP. If you have a specific use case, please let us know and we'll help you find the right solution faster.
By submitting this form you are agreeing to our Privacy Policy and Website Terms of Use.
Related News
Five reasons why you need Microsoft 365 services
Microsoft 365 is the engine room of the modern workplace – but only when it’s configured to fit your business needs. Airtight security. Effortless collaboration. AI built for work. All of this is at your fingertips with Microsoft 365. But whether your...
Adobe Creative Cloud for Teams is retiring for Education
Adobe Creative Cloud for Teams retires for higher education on 31 March 2026.
This is your chance to modernise creativity, collaboration, and compliance across campus. Explore Creative Cloud Pro Plus (Enterprise) and plan a smooth migration that empowers teaching, learning, and research.
We’re exhibiting at Cloud & AI Infrastructure London 2026
4 - 5 March 2026 9:00 am - 5:00 pm GMT
We’re excited to be heading back to London for this year’s Cloud & AI Infrastructure London 2026. Make sure to grab your ticket and stop by our stand D212. We’d love to hear your updates on what’s new in your organisation – the plans and challenges you’re facing in 2026. ...
Cyber security in 2025: What we learned and how to stay ahead in 2026
Why human factors, AI, and compliance will define the next era of cyber defence 2025 was a turning point for cyber security. From headline-grabbing breaches to the rise of AI-driven attacks, businesses faced unprecedented challenges. So, what can you learn – and how can...