TornadoVM v2.0.0 release with lots of improvements for Java on GPUs (github.com)

The release of TornadoVM v2.0.0 introduces significant enhancements for Java applications running on GPUs. Key improvements include support for GPU-native Int8 types, streamlined packaging as an SDK across multiple platforms, and implementation of zero-copy memory management, which enhances efficiency for developers. Additionally, this version supports various data types in local memory, such as byte and half-float arrays, and allows for FP32 to FP16 conversion, providing greater flexibility for performance optimization and resource management on different hardware accelerators. This release is particularly noteworthy for the AI/ML community as it simplifies the process of harnessing GPU capabilities in Java, enabling faster computation for machine learning tasks. The updates also improve compatibility with various backends and address critical bugs that could hinder performance. With infrastructure enhancements aimed at easing unit testing and deployment to Maven Central, TornadoVM v2.0.0 sets a solid foundation for future advancements, ensuring developers can efficiently utilize GPU resources in their AI and machine learning applications.