Intel® oneAPI Base Toolkit Release Notes

Bug fixes in Compilers, oneCCL and PTI, aimed at improving performance and developer productivity.

Toolkit Level Updates

• Improved cross-platform compatibility, performance, and efficiency with the Intel® oneAPI DPC++/C++ Compiler's enhanced SYCL interoperability with Vulkan* and DirectX* graphics APIs, benefiting applications like rendering and data visualization, among many others.
• C++ developers gain more efficient parallel algorithms programming coding options using Intel® oneAPI DPC++ Library (oneDPL) expanded range-based APIs according to the upcoming C++26 standard, as well as the oneDPL set algorithms performance enhancement
• Data center application developers can now use the combined power of multiple GPUs for 2D and 3D non-batch FFTs using the new SYCL DFT API in Intel® oneAPI Math Kernel Library (oneMKL).
• Easily migrate CUDA code to SYCL with the Intel® DPC++ Compatibility tool automatically migrating 357 more APIs used by popular AI and accelerated computing apps.
• Streamline your debugging efficiency in Visual Studio with advanced thread filtering, focusing on active threads and simplifying error management, ideal for game development and real-time simulations. Enjoy regex validation, error handling options, and visual indicators for a more efficient and intuitive debugging experience.
• Enhanced debugging experience with the ability to set breakpoints on GPU kernels written in assembly from various inferiors, ideal for code offloaded to the GPU such as AI workloads. This includes debugging library calls from libraries like oneDNN and oneMKL, ensuring a more streamlined and efficient development process.
• Users of precision imaging apps can manipulate and align 32-bit floating point image faster on Intel® Advanced Vector Extensions 512 (Intel® AVX-512) equipped Intel® CPUs, both for apps using the Intel® Integrated Performance Primitives (Intel® IPP) Warp Perspective functions with Nearest Interpolation, or apps using OpenCV built with Intel® IPP
• Users of Intel® IPP Phase function to process 16-bit-integer image/signal data get several times faster performance on AVX2 and AVX512 capable CPUs
• Providing your app users the peace of mind of future-proof data security, by using Intel Cryptography Primitives Library new XMSS key/signature generation and enhanced SHA3 algorithms to protect data against classic and quantum computing threats
• Intel Cryptography Primitive Library's new HKDF algorithms empower developers to create short and strong secret keys to protect data and secure communication
• AI developers can achieve faster performance of DirectML-based AI applications on Intel client GPUs by optimizing for performance bottlenecks.
• AI developers can now optimize their AI applications running on Intel NPUs for performance by understanding the application frequency.

Intel® oneAPI DPC++ Compiler 2025.2.0

New OpenMP 6.0 features:

• Added the stripe loop-transformation construct. Striping a loop or loop nest is useful to tune GPU offload performance.
• The nowait clause in target, target enter/exit data, and target update constructs can now take an optional Boolean argument to conditionally choose between asynchronous or synchronous offloading.

New OpenMP behaviors:

• Better support of debug info in OpenMP-outlined routines where some variables were previously reported as optimized away by gdb.
• For spir64 devices, a new command-line flag, -fopenmp-target-teams-default-vla-alloc-mode=malloc/wilocal (default: wilocal), was added to allow control over how local copies for variable-length arrays private to teams and distribute constructs are allocated. 
• Added a new command-line flag, -fopenmp-target-loop-stride=local-size/global-size/one (default: local-size), to tune performance of spir64-offloaded OpenMP loops by controlling their loop stride. 

Intel® oneAPI DPC++ Library 2022.9.0

The Intel® oneAPI DPC++ Library 2022.9.0 

• Introduced additional range-based APIs for parallel algorithms. This work continues to improve alignment of oneDPL algorithm APIs with the upcoming C++26 standard.
• Improved performance of set algorithms.

Intel® DPC++ Compatibility Tool 2025.2.0

The Intel® DPC++ Compatibility Tool 2025.2.0

• Improved API coverage by supporting migration of 49 APIs (cuBlasLt/cuSparse/nvSHMEM/Texture/Graphics interop/ASM) used by priority apps, and other 308 APIs (nvSHMEM/Graph) with the available mappings.
• Support migration of CUDA Vulkan/DX Graphics Interop for 5 semaphores and 4 APIs.

Intel® oneAPI Math Kernel Library 2025.2.0

• Introduces distributed SYCL DFT API with PVC support for computing 2/3D non-batch FFT on multiple GPUs.
• Introduced counter_engine_adaptor and (u)int8/16 types support in SYCL RNG Device API.
• The compiler team introduced device image compression in version 2025.0.1, and this feature is now enabled in oneMKL 2025.2 builds. It facilitates the reduction in size of SYCL libraries with large kernels by approximately 12% to 40% for both static and dynamic libraries on Linux, as well as dynamic libraries on Windows.

Intel® Distribution for GDB* 2025.2.0

• Adds thread filtering options into VS and VSCode IDEs.
• Enhances CLI thread filter functionality.
• Adds group thread resuming feature.
• Enhances debugging of assembly GPU Kernels for oneDNN and oneMKL.

Intel® VTune™ Profiler 2025.4.0

• Adds DirectML support via PIX markup device (GPU).
• Expands the reported set of NPU performance metrics (frequency).

Intel® Advisor 2025.2.0

• Includes security and bug fixes.

Intel® oneAPI Threading Building Blocks 2022.2.0

The Intel® oneAPI Threading Building Blocks (oneTBB) 2022.2.0

• Links tbbbind with Static HWLOC for NUMA API predictability.

Intel® Integrated Performance Primitives 2022.2.0

• Enables warp() Intel® IPP calls in OpenCV build with Intel® IPP.
• Enables Intel® AVX-512 optimization for ippiWarpPerspectiveLinear_32f_C1R. 
• Improves performance for DFT/IDFT_16fc 72 and 216.

Intel® Cryptography Primitives Library 2025.2.0

The Intel® Cryptography Primitives Library 2025.1.0 optimized SM4 and SHA-512 hash family algorithms for Lunar Lake, Arrow Lake S and Clearwater Forest CPUs (optimization is based on the new SM4 and SHA512 instructions respectively).

• Crypto Multi buffer library was extended with Intel® AVX-IFMA implementation of ECDSA (Sign and Verify), public key generation, ECDHE over NIST p256r1 curve.
• Added support for HKDF, Hashed Message Authentication Code (HMAC)-based key derivation function as defined by RFC-5869.
• Added support for SHA3-224, SHA3-256, SHA3-384, SHA3-512, SHAKE128 and SHAKE256 hash algorithms as defined by FIPS PUB 202.
• Added support of Key and signature generations for the eXtended Merkle Signature Scheme (XMSS) algorithm. 

Intel® oneAPI Collective Communications Library 2021.16.0

The Intel® oneAPI Collective Communications Library 2021.16.0 adds:

• SYCL graph support for Record and Replay for Allgather, Allreduce, and ReduceScatter.
• SYCL-based implementation of ring algorithm for Allgather.
• SYCL-based implementation for Broadcast.
• Multithread support for Allgather and ReduceScatter scale up impementation.
• Attribute in the communicator to specify blocking operations for CPU.

Intel® oneAPI Data Analytics Library 2025.6.0

The Intel® oneAPI Data Analytics Library 2025.6.0 contains:

• Better performance on linear regression with over 4k features
• Usage of higher SIMD instruction sets on AMD
• Added correlation distance metric
• Fixed a crash in GPU KMeans when the number of rows is large

Intel® oneAPI Deep Neural Network Library 2025.2.0

Intel® oneAPI Deep Neural Network Library 2025.2.0 has

• Improved performance on Lunar Lake and Arrow Lake integrated GPUs and Battlemage discrete GPUs.
• Improved performance on Granite Rapids CPUs.
• Extended flexibility of SDPA, MLP, and GQA fusion patterns with Graph API.

Bug fixes in Intel® Profiling Tools Interface (Intel® PTI)

Bug fixes in onCCL and Intel® PTI

Bug fixes in Compilers, oneCCL, oneDNN, Intel® PTI, Vtune, Advisor, and GDB

Toolkit Level Updates

• The Intel® oneAPI DPC++/C++ Compiler, which already offers CPU MemorySanitizer support, now extends this capability to the device side, including GPUs. This enhancement allows you to easily detect and troubleshoot issues in both CPU and device code, ensuring more reliable applications.
• The Intel® oneAPI DPC++/C++ Compiler now supports ccache* to significantly speed up your build times. By caching previous compilations and reusing them, developers can experience faster iterations and more efficient workflows, allowing you to focus on writing high-quality code rather than waiting for builds.
• The Intel® oneAPI DPC++/C++ Compiler's code coverage tool now includes GPU support and enhanced CPU coverage for applications using C/C++, SYCL, and OpenMP. It offers you detailed analysis and comprehensive HTML reports to identify tested and untested code sections, ultimately improving test coverage and code quality while ensuring easy integration into workflows.
• The Intel® oneAPI DPC++/C++ Compiler’s integrated support for Altera FPGA has been removed as of the 2025.1 release. Altera* will continue to provide FPGA support through their dedicated FPGA software development tools. Existing customers can continue to use the Intel® oneAPI DPC++/C++ Compiler 2025.0 release which supports FPGA development and is available through Linux* package managers such as APT, YUM/DNF, or Zypper. Additionally, customers with an active support license can access the Intel® oneAPI DPC++/C++ Compiler 2025.0 via their customer support account.For more information and assistance with transitioning to the Altera development tools, please contact your Altera representative.
• GPU kernels run up to 3x faster for algorithms including copy, transform, order-changing, generation and set operations with Intel® oneAPI DPC++ Library (oneDPL), which also adds support for copy, transform and merge range-based C++20 standard algorithms
• Developers now have more random number distribution choices with the addition of geometric distributions to the already long list supported by the Intel oneAPI Math Kernel Library.
• Developers can experience faster workload execution via Fast Fourier Transform performance improvements for certain cases on Intel discrete and integrated GPUs.
• Intel® oneAPI Threading Building Blocks (oneTBB) enables parallel processing of tensors in any dimensions, with the new blocked_nd_range feature to expand support from 3 to N dimensional
• Intel® VTune™ Profiler
  • Identify performance bottlenecks of AI workloads that are calling DirectML or WinML APIs.

  • Understand the overall accelerator performance by seeing GPU and NPU offload bottlenecks in one view. 

  • Pinpoint the most time-consuming code sections and critical code paths for Python 3.12.
• Easily migrate CUDA code to SYCL with the Intel® DPC++ Compatibility tool automatically migrating 158 more APIs used by popular AI and accelerated computing apps

• Gaming, graphics, and content creation developers and ISVs – deliver real-time visual AI and gaming experiences when using C++ with SYCL for GPU acceleration​

  • Enhanced SYCL interoperability with Vulkan and DirectX12 enables the sharing of image texture map data directly from the GPU, eliminating extra image copying between CPU and GPU, ensuring seamless performance image processing and advanced rendering applications, and boosting content creation productivity
• The Intel Distribution for GDB* rebases to GDB* 15.2 staying current and aligned with the latest enhancements supporting effective application debug.
• The Intel Distribution for GDB* now includes scheduler-locking IDE options to step by default in VSCode* when debugging on a Linux* machine, providing more precise control and a enhanced debugging experience.
• Intel® Distribution for GDB* adds support for Intel® Core™ Ultra processors (Series 2) and Intel® Arc™ B-Series Graphics on Linux*, in addition to the existing Windows support, allowing developers to efficiently debug application code on these new CPUs and GPUs.

• Boost your image processing applications with Intel® Integrated Performance Primitives' in-place MulC function, now supports image sizes up to 512 Megapixels for high-performance, low-memory overhead image scaling

• Optimize your 5G signal processing with Intel® Integrated Performance Primitives' enhanced Discrete Fourier Transform (DFT), specifically tuned for the critical demands of cellular systems

• Develop more secure and efficient applications using Intel® Cryptography Primitives library SM4 and SHA-512 hash algorithms optimized for Intel® Core Ultra 200V, Intel® Core Ultra 200S and upcoming Intel® Xeon® Scalable processor
• Developers can enhance matrix multiplication and convolution performance with oneDNN, optimized for Intel® Xeon® processors. These improvements leverage the Intel® Advanced Matrix Extensions (Intel® AMX) instruction set, making it ideal for datacenter AI workloads, ensuring your applications run efficiently on the latest Intel architectures.
• Developers running AI inference on client CPUs can unlock improved performance with oneDNN on Intel Arc Graphics. This optimization maximizes the capabilities of Core Ultra processors (Series 2) and Intel Arc B-series discrete graphics, ensuring your AI applications execute with greater speed and efficiency, enhancing your development efforts.
• Developers can optimize AI models with improved performance for Gated Multi Level Perceptron (Gated MLP) and Scaled Dot-Product Attention (SDPA) with implicit casual mask using oneDNN. Support for int8 or int4 compressed key and value through the Graph API enhances both speed and efficiency, enabling more powerful and responsive AI applications.
• Developers are able to utilize collectives in more operations with added support for Average in the Allreduce and Reduce-Scatter collectives in oneCCL.
• oneCCL gives developers more control over collective communications with extensions to the Group API supporting collective operations and a new API to split a communicator.
• Users are able to scale up more effectively with oneCCL performance optimizations for Alltoall.

• Starting with the 2025.1 release, enhanced config file to support passing arguments (e.g., key=value1,value2,value3,...) to component env/vars scriptsTo define these arguments, Create a `config.txt` file that contains the following lines with list of arguments:

  mkl=latest,lp64 ilp64ipp=excludecompiler=latest,--include-intel-llvm,vs2022ippcp=2025.0dpcpp-ct=latest,vs2017tbb=2022.1,allIn the example above:

  • The first argument for each key is mandatory, while the remaining arguments are optional.
  • The first argument specifies a version directory located at the top level of the component directory. It can also be a symbolic link.
  • Subsequent arguments are comma-separated and provide additional configuration options.

Intel® oneAPI DPC++ Compiler 2025.1.0

The Intel® oneAPI DPC++/C++ Compiler 2025.1.0

• Added SYCL interoperability with Graphics APIs DirectX* 12 and Vulkan* interop so developers can create highly efficient visual compute, media processing, and rendering applications on Intel® Graphics, enhancing both performance and developer productivity.
• Extends Memory Sanitizer capability to the device side, including GPUs. This enhancement allows to easily detect and troubleshoot issues in both CPU and device code, ensuring more reliable applications.
• Supports ccache* to significantly speed up build times. By caching previous compilations and reusing them, developers can experience faster iterations and more efficient workflows, allowing you to focus on writing high-quality code rather than waiting for builds.
• The code coverage tool now includes GPU support and enhanced CPU coverage for applications using C/C++, SYCL, and OpenMP. It offers detailed analysis and comprehensive HTML reports to identify tested and untested code sections, ultimately improving test coverage and code quality while ensuring easy integration into workflows.

Intel® oneAPI DPC++ Library 2022.8.0

The Intel® oneAPI DPC++ Library 2022.8.0 

• Improved performance of copy, transform, order-changing, generation, and set operation algorithms.
• Added range-based APIs for copy, transform, merge algorithms.

Intel® DPC++ Compatibility Tool 2025.1.0

The Intel® DPC++ Compatibility Tool 2025.1.0

• Improved API coverage by enabling migration for 158 APIs (82 cuBLAS/cuSPARSE APIs, 11 Runtime/Driver API, 6 PTX, 36 SIMD math APIs, 10 texture/surface API, 1 thrust API, 5 CUB API, 8 Graphic interop API).
• Further improved build scripts migration for CMake-based APPs and Python setup.py migration of PyTorch-based Apps.
• Supported PyTorch(C10 API specific for CUDA) migration by publishing customized migration rules.
• Further improved API mapping query with more APIs.
• Supported new telemetry lib to collect migration un-supported API and usage information of DPCT features.

Intel® oneAPI Math Kernel Library 2025.1.0

Intel® Distribution for GDB* 2025.1.0

• The Intel Distribution for GDB* rebases to GDB* 15.2 staying current and aligned with the latest enhancements supporting effective application debug.
• The Intel Distribution for GDB* now includes scheduler-locking IDE options to step by default in VSCode* when debugging on a Linux* machine, providing more precise control and a enhanced debugging experience.
• Intel® Distribution for GDB* adds support for Intel® Core™ Ultra processors (Series 2) and Intel® Arc™ B-Series Graphics on Linux*, in addition to the existing Windows support, allowing developers to efficiently debug application code on these new CPUs and GPUs.

Intel® VTune™ Profiler 2025.1.0

The Intel® VTune™ Profiler 2025.1.0 added capabilities to

• Identify performance bottlenecks of AI workloads that are calling DirectML or WinML APIs.
• Understand the overall accelerator performance by seeing GPU and NPU offload bottlenecks in one view. 
• Pinpoint the most time-consuming code sections and critical code paths for Python 3.12.

Intel® Advisor 2025.1.0

The Intel® Advisor 2025.1.0

• Enabled Intel® Xe Matrix Extensions (Intel® XMX) (systolic arrays) in GPU Roofline and GPU Instruction Mix
• Improved overall GPU profiling quality and accuracy. Reduced GPU Survey collection overhead

Intel® oneAPI Threading Building Blocks 2022.1.0

The Intel® oneAPI Threading Building Blocks (oneTBB) 2022.1.0

• Enables parallel processing of tensors in any dimensions, with the new blocked_nd_range feature to expand support from 3 to N dimensional
•  Includes experimental support of Class Template Argument Deduction (CTAD) for oneTBB blocked_nd_range , to allow the compiler to auto infer class template arguments, simplifying code and improving usability
• Provides applications with multiple parallel processing phases with the ability to use oneTBB parallel_phase API to influence CPU threading resource sharing and minimize latency within phases

Intel® Integrated Performance Primitives 2022.1.0

The Intel® Integrated Performance Primitives 2022.1.0

• Added new inplace API for MulC functionality for images
• Optimized selected sizes of DFT for signal sizes important for 5G 

Intel® Cryptography Primitives Library 2025.1.0

The Intel® Cryptography Primitives Library 2025.1.0 optimized SM4 and SHA-512 hash family algorithms for Lunar Lake, Arrow Lake S and Clearwater Forest CPUs (optimization is based on the new SM4 and SHA512 instructions respectively).

Intel® oneAPI Collective Communications Library 2021.15.0

The Intel® oneAPI Collective Communications Library 2021.15.0 added

• Support for Average in the Allreduce and Reduce-Scatter collectives.
• Extensions to the Group API supporting collective operations and a new API for splitting communicators giving developers more control over collective communications.
• Performance optimizations for scale up for Alltoall

Intel® oneAPI Data Analytics Library 2025.1.0

The Intel® oneAPI Data Analytics Library 2025.1.0 contains

• Updates to the unified API for Intel® oneAPI Data Analytics Library (oneDAL) RNG primitive functions
• Changed stream handling in the RNG clone
• Fixed issues with the oneDAL code samples (building with C++20)

Intel® oneAPI Deep Neural Network Library 2025.1.0

Intel® oneAPI Deep Neural Network Library 2025.1.0 has

• Improved matmul and convolution primitive performance on Intel Xeon processors with Intel® AMX instruction set support (formerly Sapphire Rapids and Granite Rapids)
• Improved performance for Intel Arc Graphics for Intel Core Ultra processors (Series 2) (formerly Lunar Lake) and Intel Arc B-series discrete graphics (formerly Battlemage).
• Introduced initial optimizations for Intel GPUs based on Xe3 architecture
• Improved performance of Gated Multi Level Perceptron (Gated MLP), Scaled Dot-Product Attention (SDPA) with implicit casual mask, and SDPA with int8 or int4 compressed key and value via Graph API.