During summer 2026, MCS researchers are conducting evaluations of the recent OpenMP taskgraph construct (specification 6.0, not yet implemented in any mainstream implementations). Supposedly, "taskgraph" could improve the performance of iterative applications by recording and replaying the same sequence of OpenMP tasks. The team's research explores the use in applications and optimizations in ongoing runtime system implementations. One objective is to automatically map OpenMP programs to CUDA graphs and compare the performance obtained on a variety of NVIDIA GPUs.