High Performance Programming / Quant Notes / p1153
Before the beeps and the boops, I have received a handful of DMs asking for discounted access. I have made one open for 1 week (60%):
https://www.research.hangukquant.com/d056649b
proceeding…
This week - we add to the quant trader handbook (now 1153 pages). We look at modern CPU topologies, explore concepts in synchronisation, lock-free programs and how memory barriers assert program behaviour. We look at cache-aware programming in high performance systems.
We then bring all of these tools together to discuss and implement fast inter-thread messaging protocol (disruptor pattern).
We ran our Google benchmarks, TLDR: it is fast fast, ~ 2000% faster than mutex-based queue and 300% faster than Boost’s lock-free queue.
Code is presented. By the way, we have been adding readers to the quantcplib repo. Since the last post, we added
google benchmarks
google tests
examples
Tomorrow we will add the next batch to the repo. New modules will are pending. I’ll keep going, and you keep it coming!
Here are the notes: [1153]