The OpenVX Working Group at Khronos today released two new extensions, Target Kernel and Node Command, aimed at improving computer vision and AI applications on modern heterogeneous Systems on a Chip. These extensions address limitations related to efficiently distributing computation across diverse hardware and dynamically adapting processing pipelines at runtime without disrupting data flow. According to Khronos, the extensions "streamline compute workloads" and resolve issues that have become more critical with increasingly complex SoC architectures. Both new extensions are slated to be included as core features in the upcoming OpenVX 2.0 API release.