KeyHunt-Cuda check throws errors ?

[citb0in]

Hello everybody,

can anyone run Keyhunt-Cuda in check mode and post his results, please? I just realized that the check outputs an error.

Code:

./KeyHunt -c

KeyHunt-Cuda v1.08


Checking... Secp256K1

Check Generator :OK
Check Double :OK
Check Add :OK
Check GenKey :OK
Adress : 15t3Nt1zyMETkHbjJTTshxLnqPzQvAtdCe OK!
Adress : 1BoatSLRHtKNngkdXEeobR76b53LETtpyT OK!
Adress : 1Test6BNjSJC5qwYXsjwKVLvz7DpfLehy OK!
Adress : 16S5PAsGZ8VFM1CRGGLqm37XHrp46f6CTn OK!
Adress : 1Tst2RwMxZn9cYY5mQhCdJic3JJrK7Fq7 OK!
Adress : 3CyQYcByvcWK8BkYJabBS82yDLNWt6rWSx Failed !
 16S5PAsGZ8VFM1CRGGLqm37XHrp46f6CTn
Adress : 31to1KQe67YjoDfYnwFJThsGeQcFhVDM5Q Failed !
 1CTVCniYaK1sxerz1mpVzwfm2HRWYLfeHV
Adress : bc1q6tqytpg06uhmtnhn9s4f35gkt8yya5a24dptmn Failed !
 1LDMEYi5HfGkZCJqEa4pmY3BXCirjCS5u3
Check Calc PubKey (full) 1ViViGLEawN27xRzGrEhhYPQrZiTKvKLo :OK
Check Calc PubKey (odd) 18aPiLmTow7Xgu96msrDYvSSWweCvB9oBA:OK


Checking... Int

GetBase10() Results OK
Add() Results OK : 357.143 MegaAdd/sec
Mult() Results OK : 30.395 MegaMult/sec
Div() Results OK : 12.346 MegaDiv/sec
R1=1000003D1
R2=1000007A2000E90A1
Field characteristic size: 256bits
ModInv()/ModExp() Results OK
ModInv() Edge cases Results OK
.Avg = 6.13
ModInv() Results OK : 701.034 KiloInv/sec
ModInv() cycles : 3560.44
ModSqrt() Results OK !
IntGroup.ModInv() Results OK : 9.974 MegaInv/sec
ModMulK1() Results OK : 41.672 MegaMult/sec
ModSquareK1() Results OK : 42.508 MegaSqr/sec
ModInv() Cost : 60.6 S
ModMulK1order() Results OK : 5.469 MegaMult/sec

I had discovered a similar issue with VanitySearch and we know that those tools build up on VanitySearch, if I'm not mistaken. Maybe there is a connection. In the VanitySearch thread I posted HERE my finding and how I mitigated the culprit be downgrading. In the next step I was able to pinpoint and eliminate the problem and finally use the latest version of VanitySearch. The problem was in the compiler, see HERE. The fix was to use g++-9 in the Makefile.

I used the same procedure here on KeyHunt-Cuda, that means I am explicitly using g++-9 in its Makefile. But it doesn't help, I still get those errors in the check. Any clues how to fix this? It makes no sense to me using KeyHunt-Cuda without ensuring its working fine. Looking forward to any helpful comment. Thank you.

EDIT:
I'm playing around with keyhunt-cuda and trying to solve puzzle 25 but it doesn't work. Either I'm missing something, doing something wrong, or the program doesn't work reliably. Can someone of you please check this and post the result here?

Puzzle #25

Address = 15JhYXn6Mx3oF4Y7PcTAv2wVVAuCFFQNiP
hash160 = 2f396b29b27324300d0c59b17c3abc1835bd3dbb
Pubkey = 03057fbea3a2623382628dde556b2a0698e32428d3cd225f3bd034dca82dd7455a
Range = 100000-200000
PrivKey = 1FA5EE5

I tried feeding keyhunt-cuda with all three possible files: address in clear-text, address as hash160 and address as hash160 in binary as converted by the supplied python script "addresses_to_hash160.py". I also tried using CPU only or GPU only. It makes no difference, keyhunt-cuda does NOT find the key for that puzzle, whatever I tried so far.

Code:

$ cat 25_addr.txt 
15JhYXn6Mx3oF4Y7PcTAv2wVVAuCFFQNiP

$ cat 25_addr_hash160.txt 
2f396b29b27324300d0c59b17c3abc1835bd3dbb

$ cat 25_addr_hash160.bin 
/9k)�s$0
Y�|:�5�=�

I tried following command variations:

Code:

./KeyHunt -t 0 -g --range 1000000:2000000 -m address -i 25_addr_hash160.bin

Code:

./KeyHunt -t 0 -g --range 1000000:2000000 -m address -i 25_addr_hash160.txt

Code:

./KeyHunt -t 0 -g --range 1000000:2000000 -m address -i 25_addr.txt

Code:

./KeyHunt -t 4 --range 1000000:2000000 -m address -i 25_addr_hash160.bin

Code:

./KeyHunt -t 0 -g --gpui 0 --gpux 256,256 --range 1000000:2000000 -m address -i 25_addr_hash160.bin

I even tried wider ranges, like ...

Code:

./KeyHunt -t 0 -g --range 10000:200000000 -m address -i 25_addr_hash160.bin

It works only when provided with the address in the command line, e.g:

Code:

$ ./KeyHunt -t 0 -g --gpui 0 --gpux 256,256 -m address --coin BTC --range 1000000:2000000 15JhYXn6Mx3oF4Y7PcTAv2wVVAuCFFQNiP

Conclusion --> KeyHunt-Cuda in single address mode requires an address as input, it does not work with a hash160 no matter if its binary or not.

However my first part of this post and the question "Why does Keyhunt-Cuda check throws errors", still remains.

[7isce]

Its easy error

See use "-m address" if use CPU with ONE thread only otherwise (GPU or CPU with more than one thread) use "-m addresses"

Good luck hunting

[nc50lc]

Seems like KeyHunt-Cuda only use -i in "addresses" mode not in "address" mode while it set the latter by default (if -m isn't provided).
The workaround is force it to -m addresses instead.

Here's some tests
Command1:

Code:

./keyhunt-Cuda -m addresses -i input.dat --range 1000000:2000000

Found.txt:

Code:

15JhYXn6Mx3oF4Y7PcTAv2wVVAuCFFQNiP 0000000000000000000000000000000000000000000000000000000001FA5EE5 03057FBEA3A2623382628DDE556B2A0698E32428D3CD225F3BD034DCA82DD7455A

Command2:

Code:

./keyhunt-Cuda -m address -i input.dat --range 1000000:2000000

Result:

Code:

Error: Wrong search mode provided for multiple addresses or xpoints

For the test failures, I think it's because it never supported P2SH nor Bech32.
Looks like it's the same test designed for Vanitysearch (CMIIAW).