RPNX::QueryGraph
Typed, memoized, concurrent query evaluation for C++23
Loading...
Searching...
No Matches
Todo

  • Optimize scheduler for many-CPU machines and improve fanout.
  • Add an optional estimated_memory_class to QueryGraph implementation specifications.
  • Add co_await estimated_memory_requirement(...) to allow blocking scheduling of memory-intense nodes, even when additional CPU threads are available. At least one node needs to be runnable at once, regardless of the estimates.