Closed jhawthorn closed 2 weeks ago
Previously we serialized the stack tables exactly the same way for each thread, which for threads with few unique stacks ended up being very wasteful.
In some cases this might make the gzip less effective, but should make writing the output faster and the uncompressed JSON much smaller.
I need to make sure this also replaces markers
I think I merged this separately
Previously we serialized the stack tables exactly the same way for each thread, which for threads with few unique stacks ended up being very wasteful.
In some cases this might make the gzip less effective, but should make writing the output faster and the uncompressed JSON much smaller.