We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Compilation Speedup Using C++ Modules: A Case Study - Chuanqi Xu - CppCon 2022
Learn how named C++ modules can significantly speed up compilation, with a case study demonstrating a 2.32x to 1.7x increase in compilation speed using compiler optimizations and inlining.
- Named modules can provide significant speedup in compilation.
- Inlining can help remove multiple deletions and improve compilation speed.
- Function reduction can be useful for reducing compilation time.
-
The
import extra limitoption can control the threshold of lines of instruction to import modules. - Compiler optimizations such as Intraprocedural and Interprocedural optimization (IPO) can help improve compilation speed.
- Single TU can be cheap and provide many compiler optimizations.
- Modules can help reduce the number of inline functions.
- The complexity theory can be correct but be too rough for users.
- The study only talks about named modules and does not cover other types of modules.
-
The default value for
import extra limitis 100. - The study shows that modules can provide a speedup of 2.32x for demo examples and 1.7x for tests.
- The study uses a case study of a demo example to demonstrate the improvement in compilation speed.
- The study also provides possible improvements in the future such as improving compiler side optimizations.
- The study uses a traditional theory to explain why modules can be faster in compilation.