Messaging System Performance Tests
This page is an entry point to published messaging-performance benchmarks, not a place to state unconditional performance conclusions without the test context.
The currently public benchmark repository is:
When reading those results, check at least:
- which system and scheduler conditions were used;
- whether the path being measured is in-process
Topic, Linux shared memory, or something else; - the payload size, subscription form, publish frequency, and statistics method.
If you first want to understand which mainline mechanisms are relevant, start from:
If you are going to write a new performance claim, attach the code path, input scale, and measurement method directly instead of relying on short phrases such as "sub-microsecond" or "faster than std::memcpy" without context.