Hi, I'm a complete Artiq amateur so please excuse my inevitable fundamental misunderstandings.
I want to save TTL input data to datasets as close to live as possible when running a repetitious experiment, but using either append_to_dataset or mutate_dataset requires a delay of around 100us to avoid underflow errors, which may be a lot for our purposes. We can wait to send the collected data to datasets at the end of the experiment and still have it run fine, though we'd still like to minimize the time cost as much as possible. Using an async RPC doesn't seem to help at all (I don't understand how they work at all so that's not surprising), and our tests are only running about a kilohertz with very small dataset values, so it's probably not a bandwidth issue. What would be the preferred solution? Thanks
ps the Mattermost seems to be full