In this talk, I plan to discuss TCMalloc, a high-performance memory allocator for C++ that we use in Google. Specifically, I’ll be covering points:
- Why the choice of memory allocator impacts application performance - How we’ve optimized our implementation to reduce the “data center tax” - Where we leverage C++ language features for better memory allocation - What telemetry we can obtain from TCMalloc in production, and how we leverage it
Chris Kennelly is a Software Engineer at Google. He works on developing performance optimizations for Google's core libraries Prior to Google, he worked on low-latency networks and computational molecular dynamics.
Wednesday September 18, 2019 09:00 - 09:30 MDT
Crest 4/5