Even though, the number one priority of the project is reliability, performance is taken very seriously as well. The short-term target was to provide performance similar to other widely used file systems. The current focus is to research and approach the highest possible theoretical performance.

Many, often non-linear, factors contribute to the file system performance. In time, more and more results will be added and updated here with full technical details. File system performance will be measured in different scenarios, such as, newly formatted and aged file system, single and concurrent accesses, synthetic and real-world workloads, performance during idle, CPU bound, I/O bound and in-memory workloads.

The first, synthetic benchmark was made with bonnie++ on Linux. The newly created file systems used the same partition, and the average of three runs was calculated. In the "/sec" columns the higher values are better, and in the "%CPU" columns the lower values are better. Please note that the stable ntfs-3g driver is NOT optimized yet besides an optimized version for embedded devices which is available separately.

The benchmark used Recovery Is Possible 4.0 LiveCD, Linux kernel 2.6.22.14, NTFS-3G 1.1120.