Tests
Explore how to benchmark and compare serial and parallel CSV reading in C++17 using real file sets. Understand the impact of system caching on performance and see how parallel parsing accelerates processing, especially with large datasets.
We'll cover the following...
We can run the two versions on a set of files and compare if the changes brought any improvements in the performance.
Our applications access files, so it’s harder to make benchmark correct as we can quickly end up in the file system cache. Before major runs of applications, a tool called Use SysInternal’s RAMMap app is executed to remove files from the cache. There are also Hard Drive hardware caches which are harder to release without a system reboot.
The application runs on a 6 core/12 thread machine - i7 8700, with a fast SSD drive, Windows 10.
Mid Size Files 1k Lines 10 Files
Let’s start with 10 files, 1k lines each, files are not in ...