I achieved the same result with 12 devices in a single machine

"
KeyHunt-Cuda v1.07

COMP MODE    : COMPRESSED
COIN TYPE    : BITCOIN
SEARCH MODE  : Single Address
DEVICE       : GPU
CPU THREAD   : 0
GPU IDS      : 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11
GPU GRIDSIZE : 16384x512, 16384x512, 16384x512, 16384x512, 16384x512, 16384x512, 16384x512, 16384x512, 16384x512, 16384x512, 16384x512, 16384x512
SSE          : YES
RKEY         : 0 Mkeys
MAX FOUND    : 65536
BTC ADDRESS  : 13zb1hQbWVsc2S7ZTZnP2G4undNNpdh5so
OUTPUT FILE  : Found.txt

Start Time   : Sat Aug 12 10:07:11 2023
Global start : 2E601111111111111 (66 bit)
Global end   : 40000000000000000 (67 bit)
Global range : 119FEEEEEEEEEEEEF (65 bit)

GPU          : GPU #4 NVIDIA GeForce RTX 4090 (128x0 cores) Grid(16384x512)

GPU          : GPU #6 NVIDIA GeForce RTX 4090 (128x0 cores) Grid(16384x512)

GPU          : GPU #9 NVIDIA GeForce RTX 4090 (128x0 cores) Grid(16384x512)

GPU          : GPU #2 NVIDIA GeForce RTX 4090 (128x0 cores) Grid(16384x512)

GPU          : GPU #7 NVIDIA GeForce RTX 4090 (128x0 cores) Grid(16384x512)

GPU          : GPU #1 NVIDIA GeForce RTX 4090 (128x0 cores) Grid(16384x512)

GPU          : GPU #0 NVIDIA GeForce RTX 4090 (128x0 cores) Grid(16384x512)

GPU          : GPU #3 NVIDIA GeForce RTX 4090 (128x0 cores) Grid(16384x512)

GPU          : GPU #8 NVIDIA GeForce RTX 4090 (128x0 cores) Grid(16384x512)

GPU          : GPU #10 NVIDIA GeForce RTX 4090 (128x0 cores) Grid(16384x512)

GPU          : GPU #11 NVIDIA GeForce RTX 4090 (128x0 cores) Grid(16384x512)

GPU          : GPU #5 NVIDIA GeForce RTX 4090 (128x0 cores) Grid(16384x512)

[00:04:38] [CPU+GPU: 71921.28 Mk/s] [GPU: 71921.28 Mk/s] [C: 0.000094 %] [R: 0] [T: 19,155,554,140,160 (45 bit)] [F: 0]
"

But the time is still not adding up... if you can do the calculations and also note the costs of renting the machines. you would end up resorting to random generation of at least the 1st 1 to 3 characters because scanning all that range from 20 to wherever you think the key would be doesn't cost pennies

but the 1s and 0s scanning can assure you you scanned all the ranges covering that specifically defined bit ranges without worries
and you can just keep doing it over and over again until you find the bit configuration for the specified puzzle you're searching for

Simple combination for bits. Maybe someone can add some coding to convert 0s and 1s to hexadecimal, print in bat file, put it in bitcrack and run it.

Step 1

Let's say start scanning from 60 bits.

0s - 30 - 1s -30 = 50%
0s - 31 - 1s  29 = 51.67%
0s - 32 - 1s  28 = 53.4%
0s - 33 - 1s  27 = 55%
0s - 34 - 1s  26 = 56.67%
0s - 35 - 1s  25 = 58.34%
0s - 36 - 1s  24 = 60%

Then reverse 0s and 1s position in max 60% which is most likely where the key are.

Step 2
Convert Binary to Hexa

Step 3
Print in .CMD for continuous execution in Bitcrack.

Step 4
Run until it hits the key or else increase the percentages if it doesn't.

Make sense or any argument?

I am doing it already but doesn't have huge power like you guys. Maybe share it here and someone can crack it very soon here.

You don't write it in GPU. Just write it in Python, print it in CMD and automate it in Bitcrack. Then the whole thing transfered to GPU already with Bitcrack.

Using this code below to find the private key for puzzle 30, which specifies that the first 4 bits are 1. Running it with 12 processes on my i7-12700kf, the private key was found in about 1 minute and 28 seconds.

---

Of course, it's just trying my luck. But who isn't trying their luck for hunters who don't have a lot of gpu? I believe that any other tool used by each of us hunters with only a small amount of resources is using random mode to search for private keys, so why not rule out some combinations with lower probability to try some private keys with higher probability?

I just try it and I can find it in 30 sec on #30. For #66 it's still difficult to scan without GPU.

Try also for #35, #40, or #45.

Scan it in python without GPU too slow for 66 bit. The only way I can figure out now as I previous posted.

Step 1

Let's say start scanning from 60 bits. Use Python to do the combination.

0s - 30 - 1s -30 = 50%
0s - 31 - 1s  29 = 51.67%
0s - 32 - 1s  28 = 53.4%
0s - 33 - 1s  27 = 55%
0s - 34 - 1s  26 = 56.67%
0s - 35 - 1s  25 = 58.34%
0s - 36 - 1s  24 = 60%

Then reverse 0s and 1s position in max 60% which is most likely where the key are.

Step 2
Convert results from Binary to Hexa

Step 3
Make a batch file and run in Bitcrack to use GPU

Step 4
Run until it hits the key or else increase the percentages if it doesn't.

By using Python to eliminate those less like key first, then scan the list with GPU

If you want to play luck with #66 just change the parameter of the curve.


for example pub pk #2 (50-50 of probabilities)



or with any other pub to skip certain keys.

For example, the default parameter in the curve is this:


or in hex



If we change this to this(pub key corresponding to private key #2):


hex


our search will go from 1,2,3,4,5,6,7..... to 2,4,6,8,10.....

if we want to search for a key in the range

we divide the range by 2


if we find the pk in 1024

multiply by 2 and this will be the original pk 2048

we would just skip odds.

if you use the pubkey from pk #3 then it would be sequence 3,6,9,12....
that is, 33.3% probability