We should have the ability to directly output zstd-compressed data. Otherwise we would have to periodically kill a resolution when we run out of space, compress the save directory, then restart the computation, potentially losing a substantial amount of work. We would also need to keep the save in a clean state in the process, see #79.
I believe this would actually improve performance, because right now the files are big enough that a nontrivial amount of time is spent writing to disk.
We should have the ability to directly output zstd-compressed data. Otherwise we would have to periodically kill a resolution when we run out of space, compress the save directory, then restart the computation, potentially losing a substantial amount of work. We would also need to keep the save in a clean state in the process, see #79.
I believe this would actually improve performance, because right now the files are big enough that a nontrivial amount of time is spent writing to disk.
See #76