Intel has announced a developer kits that use the oneAPI cross-architectural programming model. The updated software tools are said to make it easier to create solutions that can speed up computing.
New additions include the world’s first unified compiler with support for C ++, SYCL and Fortran, a Python CPU / GPU implementation with data parallelization, advanced modeling and accelerator performance tuning tools, and performance enhancers for AI workloads and trace rendering. rays.
The oneAPI model leverages the strengths of heterogeneous systems that include different types of processors, processor cores and coprocessors, general purpose GPUs, programmable arrays (FPGAs), and other accelerators. They can be combined into a comprehensive solution using software that allows you to select the optimal equipment for specific tasks without the overhead and technological limitations inherent in closed programming models.
Intel oneAPI 2022 toolkits provide high productivity and developer productivity with a full arsenal of modern development tools: compilers, libraries, pre-optimized frameworks, analyzers and debuggers. Along with the already mentioned compiler and Python implementation, the set includes an improved Intel DPC ++ Compatibility Tool to automatically port 90% to 95% of CUDA code to SYCL / DPC ++.
Of course, the entire toolkit is optimized for the rich capabilities of 12th Gen Intel Core processors with AVX-VNNI support, next generation Intel Xeon processors with Intel Advanced Matrix Extension (Intel AMX), aka Sapphire Rapids, and graphics Xe processors.
The manufacturer notes expanded support for integrated development environments, including Microsoft Visual Studio Code, Microsoft Visual Studio 2022 and Microsoft WSL2.
The Intel oneAPI 2022 Developer Toolkits are already available for download or use in the Intel DevCloud cloud at no cost.