C++ HFT Optimizations

27 Apr 2025

[ c++  performance  design  ]

Avoid context switches and bypass the scheduler

Divide threads to three tiers

Networking kernel bypass

The most important optimization - cache warming

Nothing is optimization unless measured

Beware microbenchmarking

Sharing data between threads

Find Relaxations

Takeaways

Lock free queue types

Key design concepts

Example

https://gitlab.com/qspark-public/sclfq

Reference