Practical overview
When it fits
Use it when a BIP39 backup is incomplete and you have target information suitable for checking candidate derivations.
Who it is for
- People with a partial BIP39 wallet backup and known missing-word positions.
- Recovery users who can supply wallet derivation information and target hashes or filters.
Documented uses
- Represent missing mnemonic words with `*` and search possible candidates.
- Check candidates against exact hashes, Bloom filters or XOR filters.
- Select supported wallet target families and, where needed, override derivation routing.
- Run recovery work on one GPU or across multiple GPUs.
Requirements & limits
The repository is public and licensed under Apache-2.0. The documentation describes Windows and Linux builds and prebuilt distribution archives; a compatible NVIDIA driver is required on the target machine.
- Templates must be 3–48 words long and divisible by three.
- The recovery check depends on choosing suitable derivation lists, target family and target data.
- Build archives are separated by GPU architecture profiles; select a profile appropriate to the GPU generation.
- The documented input parser can auto-detect a matching BIP39 language wordlist and correct common typos when the intended word is clear.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Build the application on Linux or WSL
Configure and build the C++ application using the documented Linux or WSL build path.
Before you start
- A Linux or WSL environment.
- An NVIDIA GPU and compatible NVIDIA driver are required to run the CUDA application.
Steps
Open the repository in the Linux or WSL environment.
Run the command below.
make configureRun the command below.
make build
Expected outcome
The documented Linux build process configures and builds the application.
Guide limitations
- The documentation provided does not specify all software dependencies or GPU architecture requirements for this build path.
- Building the application alone does not recover a phrase; a recovery run also requires a template and suitable derivation and target data.
Prepare a file of incomplete mnemonic templates
Format known and missing mnemonic positions for file-based recovery.
Before you start
- A text file containing one mnemonic template per line.
- Known words and the positions of missing words.
Steps
Write one template per line, separating words with spaces.
Replace each unknown word with
*.Keep each phrase between 3 and 48 words, with a total length divisible by three.
Use the prepared file as the recovery input alongside the documented derivation list and target options.
Expected outcome
A template file in the input format expected by the recovery workflow.
Guide limitations
- Formatting the templates does not itself perform recovery.
- Candidate validation still requires appropriate derivation and target data.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 14 October 2026
- Original website title
- XopMC/CUDA_Mnemonic_Recovery · Official homepage title · retained in its source language