Intel OneAPI HPC Toolkit
Enhanced Instruments for High-Performance Computers
The foundation of AI, machine learning, and deep learning applications is high-performance computing (HPC). With the newest methods in vectorization, multithreading, multi-node parallelization, and memory optimization, the Intel oneAPI HPC Toolkit gives developers the tools they need to create, evaluate, optimize, and scale HPC programs.
This toolkit contains sophisticated analysis tools and strong data-centric libraries.
Features
Build
Intel’s industry-leading compiler technology and libraries make it easier to deploy HPC applications on CPUs and accelerators.
Analyze
Assess your application’s performance quickly, see how resource usage affects your code, and identify areas for optimization to guarantee better cross-architecture performance.
Scale
Use the standards-driven Intel MPI Library and benchmarks, MPI analyzer, and cluster tuning tools that are provided to deploy apps and solutions across shared memory and distributed memory (like clusters) computing platforms.
What’s Included
Compile and optimize conventional Fortran code for CPU and GPU architectures with the Intel Fortran Compiler.
- Provide scalable, adaptable, and efficient cluster messaging across a range of network fabrics and CPU and GPU architectures with the Intel MPI Library.
- Compile and optimize standard C++ and SYCL code for CPU, GPU, and FPGA target architectures with the Intel oneAPI DPC++/C++ Compiler.
- Intel DPC++ Compatibility Tool: Use this helper to convert legacy CUDA code to open multiarchitecture SYCL code.
- Intel VTune Profiler: Identify and enhance CPU, GPU, and FPGA system performance bottlenecks.
- Intel Advisor: Write code that is optimized for threading, offloading to accelerators, and vectorization.
- Deep, system-wide debugging of SYCL, C, C++, and Fortran programs is made possible by the Intel Distribution for GDB.
- Intel oneAPI DPC++ Library (oneDPL): Utilize these essential productivity techniques and features to accelerate data parallel tasks.
- Intel oneAPI Threading Building Blocks (oneTBB): This sophisticated threading and memory-management template library makes parallelism easier.
- Fast Fourier transforms (FFT), vector math, and matrix algebra are among the math processing operations that are accelerated by the Intel oneAPI Math Kernel Library (oneMKL).
- Boost the efficiency of photography, signal processing, data compression, cryptography, and other areas with Intel Integrated Performance Primitives.
- The Intel Cryptography Primitives Library is a collection of safe, quick, and portable cryptography building blocks that are tailored for Intel CPUs.
- Improve the performance of data analytics and machine learning using the Intel oneAPI Data Analytics Library (oneDAL).
- Create quick neural networks on Intel CPUs and GPUs using performance-optimized building elements with the Intel oneAPI Deep Neural Network Library (oneDNN).
- Distribute deep learning model training across several nodes by implementing optimized communication patterns using the Intel oneAPI Collective Communications Library (oneCCL).
- Intel oneAPI DCP++/C++ Compiler FPGA Support Package (requires separate download)
Use program FPGA accelerators to speed up specialized, data-centric workloads or SYCL high-level synthesis (HLS) to accelerate your register transfer level (RTL) development. Installing the Intel oneAPI Base Toolkit (Base Kit) is necessary for this.
Bundles
Download smaller bundles for particular development use cases in addition to the complete toolkit.
Intel C++ Essentials
The most widely used performance libraries in the Base Kit for Intel CPUs and GPUs can be compiled, debugged, and used by C++ and SYCL developers with this bundle. It consists of:
- GDB Intel Distribution
- Compatibility Tool for
- Intel DPC++
- Intel oneAPI DPC++/C++
- Compiler oneDPL
- oneMKL
- oneTBB
Intel Fortran Essentials
The most widely used performance libraries in the Intel oneAPI HPC Toolkit for Intel CPUs and GPUs can be compiled, debugged, and used by Fortran developers working on numerical tasks.
- Intel Distribution for GDB
- Intel Fortran Compiler
- OneMKL
- Intel MPI Library
Intel oneAPI HPC Toolkit Specifications
| Category | Specifications |
|---|---|
| CPUs | – Intel Xeon processor family |
| – Intel Core processor family | |
| – Other processors compatible with Intel 64 architecture | |
| GPUs | – Intel UHD Graphics for 11th generation Intel processors or newer |
| – Intel Iris Xe graphics | |
| – Intel Arc graphics | |
| – Intel Server GPU | |
| – Intel Data Center GPU Flex Series | |
| – Intel Data Center GPU Max Series | |
| – NVIDIA or AMD GPUs using plug-ins from Codeplay | |
| Languages | – C++ with SYCL |
| – Fortran (Requires Microsoft Visual Studio* on Windows) |
