ddup (Detect Duplicates) is an extremely fast tool that identifies potentially duplicated files in Windows NTFS Volumes.

Note that since the NTFS Journal is limited in size, not all duplicated files will be found.

ddup C:

ddup C: -m **\*.dmp -i

Output:

Scanning drive C: with matcher `**\*.dmp` (case-sensitive) [Fuzzy comparison]
[1/3] Generating recursive dirlist
Finished in 7.798245 seconds
[2/3] Grouping by file size
Finished in 0.0028928 seconds
[3/3] Grouping by hash in thread pool
Potential duplicates [84654 bytes]
	C:\Windows\LiveKernelReports\NDIS-20190504-0002.dmp
	C:\ProgramData\Microsoft\Windows\Containers\Dumps\f9292c13-143c-4070-98b5-7e12e2afddfc.dmp
Finished in 0.001117 seconds
Overall finished in 7.857446 seconds

Install from crates.io:

cargo install ddup

Install from repository:

cargo install --git https://github.com/netaneld122/ddup

This tool is written in Rust .

ddup obtains a recursive dirlist by leveraging the NTFS USN Journal mechanism in order to read USN records for MFT (Master File Table) entries.

Windows USN Records can be fetched via the following IOCTLs:

• FSCTL_ENUM_USN_DATA
• FSCTL_QUERY_USN_JOURNAL

The USN records represent either Files or Directories, linking one to another, so in order to resolve the full path
of a file, an SQL-equivalent "recursive join" has to be performed on the records (implemented via a HashMap).

After the full paths are resolved, we start comparing the files by using several iterations:

• Find groups of files that have the same size
• Compare files using fuzzy hashing on all cores simultaneously

The results are most probably identical, although it is not strictly guaranteed.
To guarantee total equivalence, use the --strict flag (however this may impact performance greatly)

Note that due to the implementation's nature, ddup requires elevated Administrator privileges.