An exhaustive review of userspace performance of C, C++ and Assembly code. The author collected an impresive amount of information, structured it and put it in a set of 5 free PDFs.
It features original research, like reverse engineered branch prediction algorithm for various CPUs and optimisation performed by various compilers. It also provides high-level guidance on how to design fast C++.
- Topic covered include, for various x86&ARM CPUs under Linux, BSD, Windows & MacOS:
- Branch prediction algorithms
- Cache behaviour
- Executable file formats (ELF, PXE, Mach-O)
- Memory mapping