Pollard's kangaroo ECDLP solver

[Jean_Luc]

Hi,

Yes there is a bug with suggestedDP when log2(nbKangaroo) > log(sqrt(N)). In the case above the suggested DP should be 0.

As said before, to avoid a large DP overhead, it is important to choose DP such as nbKangaroo.2^dpbit << sqrt(N) and here you have nbKangaroo.2^dpbit >> sqrt(N) !
So for 40 bits search, 2^23.9 kangaroo is too much.
As said in the readme "Powerfull GPUs with large number of cores won't be very efficient on small range", so to solve 40bit search, use only the CPU even with -t 1 (1 thread is enough), it will solve 40bit key in few seconds.

[WanderingPhilospher]

Hi,

Yes there is a bug with suggestedDP when log2(nbKangaroo) > log(sqrt(N)). In the case above the suggested DP should be 0.

As said before, to avoid a large DP overhead, it is important to choose DP such as nbKangaroo.2^dpbit << sqrt(N) and here you have nbKangaroo.2^dpbit >> sqrt(N) !
So for 40 bits search, 2^23.9 kangaroo is too much.
As said in the readme "Powerfull GPUs with large number of cores won't be very efficient on small range", so to solve 40bit search, use only the CPU even with -t 1 (1 thread is enough), it will solve 40bit key in few seconds.

COBRAS was initially running an 85 bit scan. I ran the 40 and 50 bit scans to show him that the program actually worked no matter start and stop ranges. But yes, he ran my 40 bit test with too much power and figured out the dp. However, the most important part of the tests and lessons is that this program works as designed. That was my intent.

[WanderingPhilospher]

@JeanLuc @Zielar

If you fellas haven't started searching the 120 range for the #120 puzzle, don't start now. It's too late, me and my MOAP...the Mother of All Processors are running things and will find it.

Mind as well tweak VanitySearch and try for #64     

My MOAP is dropping bombs like a rolling thunder...

[Jean_Luc]

What about your issue with server which drops client ?
Did you remove the VALIDITY_CHECK ? Is is better ?

[WanderingPhilospher]

What about your issue with server which drops client ?
Did you remove the VALIDITY_CHECK ? Is is better ?

Yeah about that...I was actually running a test with a larger -dp 40 or so, not a smaller one.  So I thought maybe it was a timeout issue; if server didn't receive anything from client it assumed the client was offline or something?

[Jean_Luc]

Ah Ok, the timeout for idle client is 1hour, after the connection is closed. However when the client will send new DP, the connection is restored.

[WanderingPhilospher]

Ah Ok, the timeout for idle client is 1hour, after the connection is closed. However when the client will send new DP, the connection is restored.

Ok, so it doesn't lose any data/dps...that's good to know.

I'm running at -dp 28 - did you all make a decision on 26 or 27 yet? I don't have all machines running for #120 so I would say you have a chance but the MOAP is the equalizer - it's able to do scalar multiplication and add P to P and 2Q to P times Pi and sqrt the order with 1,798 teraflops\s.  

edit: I am also refurbishing my i7 920 SLBCH; equipping it with some HBM2 memory and adding 28 GB cache. When that is done...look out!

[BeatrixP]

Very new to this but interesting project. I am not very good at programming or understand how this works very much but can you start at say 30% of a range as opposed to the beginning of the range? does it really matter where you start if so? Does there need to be x amount of calculations before 2 kangaroos meet? Sorry for all questions i am very interested in how this all works. I do not have a good machine so no point me trying but could someone theoretically get lucky before someone with monster gpus ?

[Shadow86]

@JeanLuc @Zielar

If you fellas haven't started searching the 120 range for the #120 puzzle, don't start now. It's too late, me and my MOAP...the Mother of All Processors are running things and will find it.

Mind as well tweak VanitySearch and try for #64     

My MOAP is dropping bombs like a rolling thunder...

I wish you a luck

Tell me is there any option to use multiply GPU on BitCrack or Vanity because i have 1k Gpu idle

[Jean_Luc]

I don't have any news from Zielar since 2  days, Maybe he has taken few deserved holidays
With the new release, we have a power of 0.5Tops

[Jean_Luc]

Tell me is there any option to use multiply GPU on BitCrack or Vanity because i have 1k Gpu idle

What kind of gpu ?

[Jean_Luc]

https://github.com/brichard19/eclambda

Can anyone try my tool on a 2080ti? On a 2080S it gets around 1300MKeys/sec when using 24-bit DP.

I tried your tool (DP18) on a V100.

Code:

[2020-06-22.11:35:35] [Info] DP: 0 TP: 0 853.74 Mpt/s (64 iter/s)
[2020-06-22.11:35:37] [Info] Verifying 40336 results
[2020-06-22.11:35:45] [Info] DP: 0 TP: 0 992.50 Mpt/s (75 iter/s)
[2020-06-22.11:35:48] [Info] Verifying 40362 results
[2020-06-22.11:35:55] [Info] DP: 0 TP: 0 991.18 Mpt/s (75 iter/s)

Kangaroo on a server too, configured in the same way, however, it is not clear how many kangaroo are running in parallel with your program and what grid setting is used.

Code:

GPU: GPU #0 Tesla V100-PCIE-16GB (80x64 cores) Grid(160x128) (207.0 MB used)
SolveKeyGPU Thread GPU#0: creating kangaroos...
SolveKeyGPU Thread GPU#0: 2^21.32 kangaroos [11.2s]
[2000.07 MK/s][GPU 2000.07 MK/s][Count 2^37.48][01:52][Server     OK]

[Shadow86]

Tell me is there any option to use multiply GPU on BitCrack or Vanity because i have 1k Gpu idle

What kind of gpu ?

Amd nvidia mixed from rx470 480 5700.. And nvidia from gtx1060 1070 1080 1080ti

[COBRAS]

Tell me is there any option to use multiply GPU on BitCrack or Vanity because i have 1k Gpu idle

What kind of gpu ?

Amd nvidia mixed from rx470 480 5700.. And nvidia from gtx1060 1070 1080 1080ti

This is a botnet ?

[Shadow86]

Tell me is there any option to use multiply GPU on BitCrack or Vanity because i have 1k Gpu idle

What kind of gpu ?

Amd nvidia mixed from rx470 480 5700.. And nvidia from gtx1060 1070 1080 1080ti

This is a botnet ?

Nope earned via crypto in old good days 

[COBRAS]

Tell me is there any option to use multiply GPU on BitCrack or Vanity because i have 1k Gpu idle

What kind of gpu ?

Amd nvidia mixed from rx470 480 5700.. And nvidia from gtx1060 1070 1080 1080ti

This is a botnet ?

Nope earned via crypto in old good days  

Okey.  

I think with so many GPU for you not needed someone else, in your plase i will be make "one man show"

p.s. For ex. For your 1000GPU Run your Kangaroo,exe server and connect all yours GPU to server. Read readme on https://github.com/JeanLucPons/Kangaroo

I wish you good luck.

[math09183]

\Tell me is there any option to use multiply GPU on BitCrack or Vanity because i have 1k Gpu idle

AFAIK on BitCrack you may use only one GPU per process, which means that yuo must share your work between different processes.

[COBRAS]

\Tell me is there any option to use multiply GPU on BitCrack or Vanity because i have 1k Gpu idle

AFAIK on BitCrack you may use only one GPU per process, which means that yuo must share your work between different processes.

I think bitcrack get some GPU power from all clients !!! 50% I think. Bitcrack online version is not Yours online, bat someone else  

[COBRAS]

Jean_Luc, why power downgrade from 8000 mkey to 1600  Mkey's ?

I run 89 Byte range on 4x2080 Ti. I have 30 GB free memory, and 34 GB used memory by Kangaroo.


start work :

Code:

Start:10000000000
Stop :1FFFFFFFFFFFFFFFFFFFFFF
Keys :22
Number of CPU thread: 5
Range width: 2^89
Jump Avg distance: 2^44.00
Number of kangaroos: 2^24.09
Suggested DP: 17
Expected operations: 2^45.57
Expected RAM: 60622.8MB
DP size: 15 [0xFFFE000000000000]
SolveKeyCPU Thread 0: 1024 kangaroos
SolveKeyCPU Thread 1: 1024 kangaroos
SolveKeyCPU Thread 2: 1024 kangaroos
SolveKeyCPU Thread 3: 1024 kangaroos
SolveKeyCPU Thread 4: 1024 kangaroos
GPU: GPU #3 GeForce RTX 2080 Ti (68x64 cores) Grid(136x256) (347.0 MB used)
SolveKeyGPU Thread GPU#3: creating kangaroos...
GPU: GPU #1 GeForce RTX 2080 Ti (68x64 cores) Grid(136x256) (347.0 MB used)
SolveKeyGPU Thread GPU#1: creating kangaroos...
GPU: GPU #0 GeForce RTX 2080 Ti (68x64 cores) Grid(136x256) (347.0 MB used)
SolveKeyGPU Thread GPU#0: creating kangaroos...
GPU: GPU #2 GeForce RTX 2080 Ti (68x64 cores) Grid(136x256) (347.0 MB used)
SolveKeyGPU Thread GPU#2: creating kangaroos...
SolveKeyGPU Thread GPU#0: 2^22.09 kangaroos [26.4s]
SolveKeyGPU Thread GPU#3: 2^22.09 kangaroos [26.5s]
SolveKeyGPU Thread GPU#2: 2^22.09 kangaroos [29.4s]
SolveKeyGPU Thread GPU#1: 2^22.09 kangaroos [29.9s]
[7319.21 MK/s][GPU 4249.73 MK/s][Count 2^34.14][Dead 0][02s (Avg 01:58:33)][19.6
[9481.52 MK/s][GPU 6403.06 MK/s][Count 2^34.79][Dead 0][04s (Avg 01:31:31)][29.6
[8483.26 MK/s][GPU 6422.12 MK/s][Count 2^35.24][Dead 0][07s (Avg 01:42:17)][39.7
[7186.98 MK/s][GPU 5890.89 MK/s][Count 2^35.58][Dead 0][09s (Avg 02:00:44)][49.7
[6349.58 MK/s][GPU 5462.11 MK/s][Count 2^35.85][Dead 0][12s (Avg 02:16:39)][59.6
[5867.72 MK/s][GPU 5193.33 MK/s][Count 2^36.08][Dead 0][14s (Avg 02:27:52)][69.6
[5574.25 MK/s][GPU 5024.18 MK/s][Count 2^36.28][Dead 0][17s (Avg 02:35:40)][79.7
[5381.29 MK/s][GPU 4913.37 MK/s][Count 2^36.46][Dead 0][19s (Avg 02:41:15)][89.8
/124.6MB]  ^CЗaвepшить выпoлнeниe пaкeтнoгo фaйлa [Y(дa)/N(нeт)]?

.....

midle of work:

Code:

[366.98 MK/s][GPU 366.71 MK/s][Count 2^44.10][Dead 0][02:00:59 (Avg 1.6d)][17582
[363.40 MK/s][GPU 363.14 MK/s][Count 2^44.10][Dead 0][02:01:01 (Avg 1.7d)][17583
[368.80 MK/s][GPU 368.53 MK/s][Count 2^44.10][Dead 0][02:01:18 (Avg 1.6d)][17588
[373.03 MK/s][GPU 372.76 MK/s][Count 2^44.10][Dead 0][02:01:21 (Avg 1.6d)][17589
[369.99 MK/s][GPU 369.73 MK/s][Count 2^44.10][Dead 0][02:01:23 (Avg 1.6d)][17590
[369.60 MK/s][GPU 369.34 MK/s][Count 2^44.10][Dead 0][02:01:26 (Avg 1.6d)][17591
.0/21995.3MB]

Jean_Luc, fix this please ? How to use this ? Avg move from 3h to 1.d !!!!

[Jean_Luc]

You are using a too small DP, too much DPs enter in the HashTable and slow down the search.
The expected RAM to reach 50% probability is 60GB.
And last but not least, you CPU work also and prevent the other threads to work.
Try DP17 or 18 and use -t 0 (no CPU) option.

Kangaroo.exe -d 17 -t 0 -gpu -gpuId 0,1,2,3 -g 136,128,136,128,136,128,136,128 in.txt