There are some performance issues with the parallel runtime at the moment. It degrades for 1) longer chains of workflows, and 2) for cyclic workflows. This (1-2) is likely due to atoms not being aligned and reconstructed when coming from different partitions. Additionally, it has poor performance when comparing running the parallel runtime with 1 partition compared to the test runtime (which in its core is very similar). This is to further explore, and suggest solutions to it.
There are some performance issues with the parallel runtime at the moment. It degrades for 1) longer chains of workflows, and 2) for cyclic workflows. This (1-2) is likely due to atoms not being aligned and reconstructed when coming from different partitions. Additionally, it has poor performance when comparing running the parallel runtime with 1 partition compared to the test runtime (which in its core is very similar). This is to further explore, and suggest solutions to it.