btcrecover
btcrecover copied to clipboard
Incorporate Improvements from John Cantrell's OpenCL Kernels (BIP39 seeds)
From here: https://medium.com/@johncantrell97/how-i-checked-over-1-trillion-mnemonics-in-30-hours-to-win-a-bitcoin-635fe051a752
Basically the ongoing task will be to look at how best to incorporate this into BTCRecover. This is likely to involve at least.
-
[ ] Add ability to basically use the existing kernel as-is for simple scenarios. (Missing or Scrambled words)
-
[ ] Add some more suitable seed generators for Seed Recovery (The existing ones were primarily desiged for string-based password recovery, this wasn't too bad when the bottlekneck was hashing on the CPU, but even the existing OpenCL stuff is bottleknecking here very badly) This will likely include both improving CPU based generation and using his GPU based work.
-
[ ] Shift more of the verification process in to OpenCL. (This should work well for just about everything other than AddressDB based tests, but even then it could be worth compressing the AddressDB list down more so GPU memory isn't an issue and just accepting that OpenCL checking will have a few false positives (even a few %, so possibly millions) which are manually re-checked on the CPU later...
-
[ ] Expand this to work with Electrum seeds too...