mormj
mormj
Closing due to lack of response.
Can you start a checklist here of blocks that would be good candidates?
Not merging until QA is properly integrated
C++ 20 has std::atomic_flag with waiting methods - if we can move to that there should be a performance gain. See: https://modernescpp.com/index.php/performancecomparison-of-condition-variables-and-atomics-in-c-20
https://github.com/facebook/folly/blob/main/folly/MPMCQueue.h
Might not require a separate scheduler, just a tracking and limiting of the "total in-flight" samples
Right now, if all blocks are placed in a single thread, the latency can be managed with set_max_fill, see https://github.com/mormj/ns-bench/blob/main/blocklib/bench/test/test_latency.py
Perhaps connect these types of blocks as normal, but have them marked as "inplace" or "nocopy", then lump them with the previous block when the buffers get created
What are the implications on the buffer object if the buffer reader and buffer writer have different sizes?
This shouldn't be too complicated: 1) store item_size of the dest port with the buffer reader 2) allow ports to be connected where overall size mismatches, but are of the...