Interview with
summary
Describes how Disruptor open-sourced led to concrete community fixes and usability changes that were reabsorbed into the codebase, improving edge-case correctness and performance; quantifies the hardware impact with the claim that a single cache miss ≈500 lost CPU instructions (worst cases up to ~1,200), so layout and access patterns matter; shows practical techniques: direct ByteBuffer off-heap and Unsafe for compact arrays, memory mapping and transferTo to avoid copies, and IPC approaches that reach tens of millions of messages per second.
AI-written summary · about
chapters (11)
- 0:00Introduction to speaking
- 1:24Open source and recruitment
- 2:08Benefits of open collaboration
- 3:58The value of external views
- 6:05Understanding mechanical sympathy
- 8:03Focusing on business problems
- 9:24Memory access and performance
- 11:51Data structure layout in Java
- 13:54Professional engineering habits
- 17:10Professional rigor in software
- 20:17Optimizing with byte buffers
related talks
45mThe Joy of Building Large Scale Systems
26mWhen Idioms Become Bottlenecks
51mDesigning for Performance
01h 04mWe Live in Interesting Times
01h 04mUnlocking Performance Through Reverse Engineering
42mAdventures in Efficiency and Performance
14mFreedom from memory interference
01h 50mUnlocking Modern CPU Power - Next-Gen C++ Optimization Techniques