https://github.com/room101-dev/Grand-Ultimate-BTC-Hacker/projects?query=is%3Aopen&type=classicNew hardware usb random generators are showing great accomplishments in using these monster bloom-filter baby-step/giant-step hacking system where on each cycle all the entire known btc high value addresses ( all 300M ) are compared on each cycle;
Problem in the past has been the weak pseudo random number generators for 'first guess', but now using /dev/random USB HW devices that cost less that $40 and or can be made for penny's using avalanche diodes on USB chips can lead ML system to first guess address to private-key translations and then iterative searches
Given 1,000's of private keys per hours on each RTX-3080 ( 5,000 Billion * 300M BTC address compares per cycle ), even old 2gb GPU nvidia cards can do a 300M * 300M comps which is close to a trillion comps on each cycle, not uncommon to find one lost bitcoin address pair per day on a common GPU rack;