Lecture 6: speculation and performance
September 13, 2018
We discuss the full single-threaded processing pipeline, then study the performance impact of speculation on one small program. Finally, we use mmap() to improve the throughput of our running example "closest", by caching some intermediate work.