The 2024 updates add Intel GPU support to their Python distribution, support for Microsoft's open-source MLNet framework to oneAPI, more performance optimizations to their Deep Neural Network Library, improved hardware resource utilization for their Collective Communications Library, and more.
There is also expanded CUDA-to-SYCL migration, various AI acceleration improvements, the oneMKL math library integrates RNG offload, support for profiling code offloaded to NPUs, and in preview is C++ parallel STL support for easy GPU offloading.
Downloads and more details on Monday's Intel Software Developer Toolkits 2024 release via Intel.com . For those that missed the September announcement, Intel's oneAPI is also evolving into the Unified Acceleration Foundation "UXL" .