sam.gas

Got inspired by a post about the advantages of AI for new generations. AI's got flaws, and bad actors are adapting. So, which field is better for business? AI or Cybersecurity?

Yeah, but there’s a catch. If you don’t know what you’re doing and don’t cool your CPU well, it can overheat.

No problem! Glad you’re sharing it.

Yeah, I’ve tried pretty much every BTC address out there. Seems like everything with known flaws has already been compromised.

Not from 2011, but there's this site from 2013 to 2015 you could check out. It's archived but might help.

This sounds kinda like what you’ve got on GitHub. Been doing experiments for years now and still can’t figure it out. Tried countless scripts and different RNGs from here.

So, I’ve been messing around with timestamps for the 32 BTC Puzzle, trying to replicate it from 2010 to 2015. Tried everything from January 1, 2010, to January 15, 2015 but no luck. It looks like this puzzle wasn’t made…

Sounds like a mishap during copy/pasting! You probably missed the <array> header. Here’s the corrected version again: https://github.com/NoMachine1/Cyclone I’m hitting about ~380M keys/s here.

You can use files from Cyclone’s github, the script will work with it.

AMD Ryzen Threadripper PRO 7985WX. But it’s not just puzzles, I'm doing a lot more with it.

I’m getting around ~190 Mkey/s directly searching Hash160 targets in pure random mode. https://github.com/NoMachine1/Cyclone

If you wanna see notable speed ups, implementing the whole SECP256K1 curve from scratch using AVX2 would be essential. Could mean 2x to 5x faster. Plus, assembly could help even more. Just optimizing hashing alone won't…

Hey! Nothing really beats this: https://github.com/JayDDee/cpuminer-opt/tree/master/algo/ripemd (ripemd) https://github.com/JayDDee/cpuminer-opt/tree/master/algo/sha (sha) It supports 4-way, 8-way, avx2/avx512.... Been…

Definitely! The inversion tree is parallelized, allowing SIMD optimizations. Field multiplications can run in parallel, making it quicker to create and break down the inverses.

You could set up a Memory-Mapped DP System with a Bloom filter in RAM and a memory-mapped DP table on disk. Just be ready for the massive size of those tables.

Cool but where's the part for loading DP.bin with load-dp coming from? I added a new command-line option to specify a DP.bin file. If you use load-dp, it'll load from there instead of generating. Plus, everything else…