[bob1029]

I spent a solid few days chasing this exact damn rabbit.

I thought I could beat the PQ implementation in .NET with something like this but I never even got close.

I think my use case breaks the assumptions in this paper due to the volatility of the distribution over time.

Edit: For reference, this is the approach taken by .NET - https://en.m.wikipedia.org/wiki/D-ary_heap

[neonsunset]

If you would like to contribute, there might be a better optimization opportunity in the current bounded Channel<T> implementation: https://github.com/dotnet/runtime/discussions/104791#discuss...