Intel has released its oneAPI 2022 toolkits to the general public. These toolkits feature expanded cross-architecture features including open-source software components.
The latest update, tagged 2021.4, is the fourth release of oneAPI tools. The toolkit contains multiple tools that Intel says are purpose-built for developers. Needless to mention, these APIs primarily help developers extract the best possible performance from Intel CPUs and GPUs/XPUs. Support for some of the upcoming flagship CPUs, including Xeon Sapphire Rapids, are already included.
The Intel oneAPI 2022 updates the LLVM-based components. The company has also included its latest work on GPU/XPU offloading, a new Intel extension for SciKit-Learn, the Intel Neural Compressor, and more. According to Intel’s Press Release, the latest update contains:
“New capabilities include the world’s first unified compiler implementing C++, SYCL and Fortran, data parallel Python for CPUs and GPUs, advanced accelerator performance modeling and tuning, and performance acceleration for AI and ray tracing visualization workloads.”
Intel says that the developments to the oneAPI platform will help developers as 40 percent of them “target heterogeneous systems that use more than one type of processor, processor core or coprocessor”, quoting an Evans Data survey.
📢Hey developers! Intel #oneAPI 2022 toolkits are now available.
— Intel News (@intelnews) December 22, 2021
Updated toolkits significantly expand oneAPI’s scope of cross-architecture capabilities available to #developers to create what’s next.https://t.co/I7gwNiHGav
With the latest update, Intel claims to have created the world’s first unified compiler implementing C++, SYCL and Fortran for CPUs and GPUs utilizing a common LLVM backend. The 2022 Intel oneAPI toolkit contains a complete set of advanced tools including compilers, libraries, pre-optimized frameworks, analyzers, and debuggers, notes Intel.
The CPU maker claims the oneAPI can accelerate deep learning framework performance by up to 10 times, and speed up machine learning algorithms by more than 100 times on Intel CPUs over the stock open-source version.
The 2022 Intel oneAPI toolkit is now available to download or use in the Intel DevCloud for free.