Solving ECDLP with Kangaroos: Part 1 + 2 + RCKangaroo

[raminnajafi]

Hi @RetiredCoder,
I tried to send you a private message, but it seems that your settings don’t allow messages from new users.
If possible, please enable messages from newbies so I can contact you.
Thank you!

[Morexl]

Please can any one tell me what is the dp size for 96 bit range ?

it's give this error "DPs buffer overflow, some points lost, increase DP value!"

[hoanghuy2912]

Can you upgrade to a more distributed system with multiple GPUs per node?

[RetiredCoder]

Hi @RetiredCoder,
I tried to send you a private message, but it seems that your settings don’t allow messages from new users.
If possible, please enable messages from newbies so I can contact you.

I don't use private messages, but you can write here.

Please can any one tell me what is the dp size for 96 bit range ?
it's give this error "DPs buffer overflow, some points lost, increase DP value!"

It depends on what you want - less DP overhead or to use less RAM, try to increase DP by 4 or so.

Can you upgrade to a more distributed system with multiple GPUs per node?

Sorry, currently I'm busy with much more interesting things (also I'm going to publish something interesting when I have time), but you can do it by yourself because RCKangaroo is open source.

[JackMazzoni]

Ok, in next version I will add support for precomputed DB to solve ranges like 80bit within seconds.

What do you mean precomputed DB? Do did you precompute the tame? Or this is like babysteps?

Does your kangaroo always move forward and does not go backward? The tame will go first then the wilds is chasing the tame? When the wilds step on the dp = solved? I cannot understand the code. I was just curious how this thing works.

[verybitcoinwow]

Many Thanks RetiredCoder.

Newbie here getting 4,375 MKeys/s with my 3090:) Will adding a 3060 to the mobo pcie 3.0 slot improve workload? Split it and require a new config? Also, since I'm learning how to do my own blind tests what ranges can you recommend for optimum results?

[snaz3d]

Hi,

Thanks for the tool, I've tested it. It seems that anytime a point resides in the second half of a range, it never gets to it. Something is wrong here!

Cheers!

[retrograded]

Hi @RetiredCoder,
I tried to send you a private message, but it seems that your settings don’t allow messages from new users.
If possible, please enable messages from newbies so I can contact you.

I don't use private messages, but you can write here.

Please can any one tell me what is the dp size for 96 bit range ?
it's give this error "DPs buffer overflow, some points lost, increase DP value!"

It depends on what you want - less DP overhead or to use less RAM, try to increase DP by 4 or so.

Can you upgrade to a more distributed system with multiple GPUs per node?

Sorry, currently I'm busy with much more interesting things (also I'm going to publish something interesting when I have time), but you can do it by yourself because RCKangaroo is open source.

 you solved puzzle 130 two years ago within 2 months, by now you prolly have more advanced hardwares, and with time much more improved algorithms. So now if you were to be given another puzzle in same range like #130, how long will it take for you to crack it? Prolly 1 Month I think?

[Torin Keepler]

Hi @RetiredCoder, I have a technical question for you. In your code, you use two wild kangaroo herds. What do you think about adding two more wild herds that would be symmetric with respect to half of the search range?

For example, suppose the range is from 0 to 10. The first wild herd starts at 7, and the second wild herd is calculated by the formula 7 − end(10) = 3. Suppose they meet at point 9: the first herd (starting at 7) has traveled a distance of 2, while the second herd (starting at 3) has traveled a distance of 6. The key could then be computed as (end + 6 − 2) / 2 = 7 or (end + 2 − 6) / 2 = 3.

Does adding new types of herds increase the probability of a collision?

[RetiredCoder]

Thanks for the tool, I've tested it. It seems that anytime a point resides in the second half of a range, it never gets to it. Something is wrong here!

Any proofs? Last guy who told me that my app does not work just missed zero in range number 

you solved puzzle 130 two years ago within 2 months, by now you prolly have more advanced hardwares, and with time much more improved algorithms. So now if you were to be given another puzzle in same range like #130, how long will it take for you to crack it? Prolly 1 Month I think?

Yeah I improved my software after solving #130, and I'm going to make it open source. 

Hi @RetiredCoder, I have a technical question for you. In your code, you use two wild kangaroo herds. What do you think about adding two more wild herds that would be symmetric with respect to half of the search range?

For example, suppose the range is from 0 to 10. The first wild herd starts at 7, and the second wild herd is calculated by the formula 7 − end(10) = 3. Suppose they meet at point 9: the first herd (starting at 7) has traveled a distance of 2, while the second herd (starting at 3) has traveled a distance of 6. The key could then be computed as (end + 6 − 2) / 2 = 7 or (end + 2 − 6) / 2 = 3.

Does adding new types of herds increase the probability of a collision?

What you say is not a new type of herd, it's the same type, just with shifted/additional range.
You can try to add/change ranges for all three types of kangs (tame, wild1, wild2) to get lower K, though I have reasons to doubt that it will help.
If you could add fourth kang type with similar range (wild3) it would help