Pollard's kangaroo ECDLP solver - page 59.

WanderingPhilospher

full member

Activity: 1162

Merit: 237

Shooters Shoot...

Quote from: NotATether on May 01, 2021, 02:52:23 PM

Quote from: WanderingPhilospher on May 01, 2021, 10:46:01 AM

I don't know anyone mass buying V100s; it is just the example used to solve #120 because that is the type and number of cards that were used to solve #110 and #115.

I feel like #120 is going to be solved by large numbers of older cards lying around in people's garages somewhere, rather than with anything that can be rented with AWS or Vast. Even 100 of any high-end card just seems too expensive for someone with deep pockets to rent in the long run and you can get better results searching with a few thousand of people's Maxwell and Pascal cards (and even Radeons if Kangaroo ever gets ported to OpenCL).

Well, it will be me so I can tell you they are all RTX 20xx and 30xx, but I am using 1 GTX 1060 and 1 GTX 1050Ti to do the tedious work. Wink

Zielar has/has access to 300+ of the V100s; and he is the one who solved 110 and 115, so that is just kind of the benchmark used to compare how long to solve 120.

EDIT: I'm not running it full time, just here and there, so anyone really has a chance.

NotATether

legendary

Activity: 1568

Merit: 6660

bitcoincleanup.com / bitmixlist.org

Quote from: WanderingPhilospher on May 01, 2021, 10:46:01 AM

I don't know anyone mass buying V100s; it is just the example used to solve #120 because that is the type and number of cards that were used to solve #110 and #115.

I feel like #120 is going to be solved by large numbers of older cards lying around in people's garages somewhere, rather than with anything that can be rented with AWS or Vast. Even 100 of any high-end card just seems too expensive for someone with deep pockets to rent in the long run and you can get better results searching with a few thousand of people's Maxwell and Pascal cards (and even Radeons if Kangaroo ever gets ported to OpenCL).

WanderingPhilospher

full member

Activity: 1162

Merit: 237

Shooters Shoot...

Quote from: NotATether on May 01, 2021, 08:48:10 AM

Quote from: bigvito19 on May 01, 2021, 08:43:20 AM

Quote from: fxsniper on April 30, 2021, 08:52:10 PM

Quote from: bigvito19 on April 26, 2021, 09:39:50 PM

256 card x Tesla V100 32GB is close to or around 300 billion keys per second. I'm going after a 256 bit key with cpu now.

Can 256 card x Tesla V100 32GB try with 64 bit puzzle #64 still can not crack in one day.

and still can not crack 120 bit puzzle #120 by kangaroo

need to discover other algorithms more

No, it would take up to 2 years or a little bit more to search the whole space of 2^64 with 256 card x Tesla V100 32GB.

Why are people still mass-buying V100s to throw at Kangaroo, when it doesn't have a deficiency like Bitcrack has?

IMO, it makes sense to splash on T4 or A100 cards which are fraction faster than the V100.

Or is it because most vendors are still stuck selling V100 GPUs and don't want to invest on newer GPUs? It's the only way you could possibly get a hold of 256 V100s in the first place.

I don't know anyone mass buying V100s; it is just the example used to solve #120 because that is the type and number of cards that were used to solve #110 and #115.

WanderingPhilospher

full member

Activity: 1162

Merit: 237

Shooters Shoot...

Quote from: bigvito19 on May 01, 2021, 08:50:52 AM

Quote from: WanderingPhilospher on April 30, 2021, 09:33:15 PM

Quote

last night I test run kangaroo python script version all night with puzzle #120 it is not work good

Did you really think you would solve it overnight running a slower python script??

For each tame, there is a wild to solve the key. If you would stop mixing hex with decimal I would try to explain it better but I'm not going to convert them back and forth. I'm still not 100% sure what you do not understand or what you are trying to figure out. But I will tell you, you may get lucky and solve the key using python script only using a CPU but the odds are low; on average it would take thousands of years to solve.

Do you know how many pairs of tames + wilds can solve a key in a given range or is it just 1 pair that can solve a key?

I'm not 100% sure but if you use DP = 0 (meaning every point would be saved) then I would say every tame has a wild that would solve the key. If one used a different DP, then it would depend on how many of those DPs exist in the range. Example, if one used DP 32, in a 2^64 range, for every tame DP of 32, then there would probably be a wild DP of 32 to solve the key.

bigvito19

full member

Activity: 706

Merit: 111

Quote from: WanderingPhilospher on April 30, 2021, 09:33:15 PM

Quote

last night I test run kangaroo python script version all night with puzzle #120 it is not work good

Did you really think you would solve it overnight running a slower python script??

For each tame, there is a wild to solve the key. If you would stop mixing hex with decimal I would try to explain it better but I'm not going to convert them back and forth. I'm still not 100% sure what you do not understand or what you are trying to figure out. But I will tell you, you may get lucky and solve the key using python script only using a CPU but the odds are low; on average it would take thousands of years to solve.

Do you know how many pairs of tames + wilds can solve a key in a given range or is it just 1 pair that can solve a key?

NotATether

legendary

Activity: 1568

Merit: 6660

bitcoincleanup.com / bitmixlist.org

Quote from: bigvito19 on May 01, 2021, 08:43:20 AM

Quote from: fxsniper on April 30, 2021, 08:52:10 PM

Quote from: bigvito19 on April 26, 2021, 09:39:50 PM

256 card x Tesla V100 32GB is close to or around 300 billion keys per second. I'm going after a 256 bit key with cpu now.

Can 256 card x Tesla V100 32GB try with 64 bit puzzle #64 still can not crack in one day.

and still can not crack 120 bit puzzle #120 by kangaroo

need to discover other algorithms more

No, it would take up to 2 years or a little bit more to search the whole space of 2^64 with 256 card x Tesla V100 32GB.

Why are people still mass-buying V100s to throw at Kangaroo, when it doesn't have a deficiency like Bitcrack has?

IMO, it makes sense to splash on T4 or A100 cards which are fraction faster than the V100.

Or is it because most vendors are still stuck selling V100 GPUs and don't want to invest on newer GPUs? It's the only way you could possibly get a hold of 256 V100s in the first place.

bigvito19

full member

Activity: 706

Merit: 111

Quote from: fxsniper on April 30, 2021, 08:52:10 PM

Quote from: bigvito19 on April 26, 2021, 09:39:50 PM

256 card x Tesla V100 32GB is close to or around 300 billion keys per second. I'm going after a 256 bit key with cpu now.

Can 256 card x Tesla V100 32GB try with 64 bit puzzle #64 still can not crack in one day.

and still can not crack 120 bit puzzle #120 by kangaroo

need to discover other algorithms more

No, it would take up to 2 years or a little bit more to search the whole space of 2^64 with 256 card x Tesla V100 32GB.

WanderingPhilospher

full member

Activity: 1162

Merit: 237

Shooters Shoot...

Quote

last night I test run kangaroo python script version all night with puzzle #120 it is not work good

Did you really think you would solve it overnight running a slower python script??

For each tame, there is a wild to solve the key. If you would stop mixing hex with decimal I would try to explain it better but I'm not going to convert them back and forth. I'm still not 100% sure what you do not understand or what you are trying to figure out. But I will tell you, you may get lucky and solve the key using python script only using a CPU but the odds are low; on average it would take thousands of years to solve.

fxsniper

member

Activity: 406

Merit: 47

last night I test run kangaroo python script version all night with puzzle #120 it is not work good

I try with puzzle #50

test 100 time, key can be found many way

Code:

Puzzle #50
611140496167764
0x22bd43c2e9354
 
the public key x-value (Hex):
f46f41027bbf44fafd6b059091b900dad41e6845b2241dc3254c7cdd3c5a16c6
110560903758971929709743161563183868968201998016819862389797221564458485814982

the public key y-value (Hex):
eb3dfcc04c320b55c529291478550be6072977c0c86603fb2e4f5283631064fb
106403041512432555215316396882584033752066537554388330180776939978150437217531

the public key (Hex):
04f46f41027bbf44fafd6b059091b900dad41e6845b2241dc3254c7cdd3c5a16c6eb3dfcc04c320b55c529291478550be6072977c0c86603fb2e4f5283631064fb

---------------
611140496167764
0x22bd43c2e9354

Collision Point (got more than in list)
a513f54f69ddf470f5abdcf769cbb6a067248d7da73197f82e07f7083eb06000 954702022026005
a513f54f69ddf470f5abdcf769cbb6a067248d7da73197f82e07f7083eb06000 343561525858241
de8298b6c0bc42029d1dceed0213ba24200d654f1ce527958555822d71f97000 1130704657170423
de8298b6c0bc42029d1dceed0213ba24200d654f1ce527958555822d71f97000 519564161002659
2bad5cd5bd5dbbfeb816a8101eb34fb64dd361388fc01162d295c3f22c819000 915124390258892
2bad5cd5bd5dbbfeb816a8101eb34fb64dd361388fc01162d295c3f22c819000 303983894091128
a8783a3787f85a8d7cec689ce70737a8da2f18a1f29543e90857910c2fdbf000 1112872798694686
a8783a3787f85a8d7cec689ce70737a8da2f18a1f29543e90857910c2fdbf000 501732302526922
8d4176027265833a19df3466b064280ad9d0487678288e1d839c0ddb90845000 857201922923546
8d4176027265833a19df3466b064280ad9d0487678288e1d839c0ddb90845000 246061426755782
b3d841d0386e6c481c151d7f218742a1c1dd4aa3ddde2a11b372926014318000 1121753485244822
b3d841d0386e6c481c151d7f218742a1c1dd4aa3ddde2a11b372926014318000 510612989077058
b3612a3e023187101fb27e0451408ea901782b07a5fc7c9f92be1f59c53d4000 1143734884281847
b3612a3e023187101fb27e0451408ea901782b07a5fc7c9f92be1f59c53d4000 532594388114083
6385520e7d0e95f03e1faf35b527d7c34476766a263a95a5b348bab526a2b000 890376348188409
6385520e7d0e95f03e1faf35b527d7c34476766a263a95a5b348bab526a2b000 279235852020645
df33e7c810e21fa26713f197679585028022b1904ec317b8281b05485152f000 1029465726095329
df33e7c810e21fa26713f197679585028022b1904ec317b8281b05485152f000 418325229927565
a6bef9a34f910573109512484fa6a979c3468df1705f264b5e21847bc70eb000 891792468051792
a6bef9a34f910573109512484fa6a979c3468df1705f264b5e21847bc70eb000 280651971884028
32db7b5f10b4c377b05354b34ca20a19f1ff9d6c73ed726e1ed33441ed0ee000 1170795663880908
32db7b5f10b4c377b05354b34ca20a19f1ff9d6c73ed726e1ed33441ed0ee000 559655167713144
c364f41146a5fef0df1c6ad295fc59158384834350e1f36e0f24719458f1d000 999241629315917
c364f41146a5fef0df1c6ad295fc59158384834350e1f36e0f24719458f1d000 388101133148153
eda7cd9b77865f998591069f115aac6cf77d3235f9c5ade22f9c044aac83a000 941328104653673
eda7cd9b77865f998591069f115aac6cf77d3235f9c5ade22f9c044aac83a000 330187608485909
bd41897bfe042ffd93c4f8284071446d8a1f47a090aefe3fe86280b209c5a000 872146232585324
bd41897bfe042ffd93c4f8284071446d8a1f47a090aefe3fe86280b209c5a000 261005736417560
b3e8568bc4eadbd87d609527d64c441241fa07bdef4e2505c4f2550585a6c000 1080535219841359
b3e8568bc4eadbd87d609527d64c441241fa07bdef4e2505c4f2550585a6c000 469394723673595
18cd29207cd04c383cf7e1a5a269539e1225b056b04a6c3bf9ca8ce18724f000 1034230412665596
18cd29207cd04c383cf7e1a5a269539e1225b056b04a6c3bf9ca8ce18724f000 423089916497832
5bb0e319c598a2543c6f96fd5c5c8f2e77ec11a2eaa188b01c5dde7bacd94000 1089872296205828
5bb0e319c598a2543c6f96fd5c5c8f2e77ec11a2eaa188b01c5dde7bacd94000 478731800038064
26374708cf690c5d95c3b0ee4e60c56760ab839d8478cff8cdc7caf54427d000 1021159839899856
26374708cf690c5d95c3b0ee4e60c56760ab839d8478cff8cdc7caf54427d000 410019343732092
5b962037e8698befb9606a0d7899c9b21e0a7e75c4dd73998dc4fa7f500b9000 886149821630225
5b962037e8698befb9606a0d7899c9b21e0a7e75c4dd73998dc4fa7f500b9000 275009325462461

What meaning each point relate with public key?

but puzzle #120 python can not run well and python can not solve

fxsniper

member

Activity: 406

Merit: 47

Quote from: bigvito19 on April 26, 2021, 09:39:50 PM

256 card x Tesla V100 32GB is close to or around 300 billion keys per second. I'm going after a 256 bit key with cpu now.

Can 256 card x Tesla V100 32GB try with 64 bit puzzle #64 still can not crack in one day.

and still can not crack 120 bit puzzle #120 by kangaroo

need to discover other algorithms more

bigvito19

full member

Activity: 706

Merit: 111

Quote from: fxsniper on April 26, 2021, 08:56:32 PM

Quote from: WanderingPhilospher on April 26, 2021, 08:28:25 PM

Range is too high and too wide for a reason...Billions of dollars in wallets/private keys. If it was too narrow of a range, no one would use Bitcoin.

It's all relative to your GPU/CPU power. You can't have one GTX 1050 and expect to find keys in 120 bit range over night, or over a decade. But if you have many GPUs, then you have a chance to find keys after time, depending on your GPU power.

challenge is too big for 256 bit it is can not possible find

challenge what maximum we can do on records, will be stuck with 120 bit forever same 256 bit or not?

if can do 120 may be can use up to 140 bit or some 2-3 level, do far from 160 bit

I think all random method is stuck on too high range, next method may be concern about calculate or mix

last records is 256 card x Tesla V100 32GB right, How many power for GPU need to solve

I try kangaroo cpu only is difference with GPU over 12-15 time, kangaroo require to use with GPU

256 card x Tesla V100 32GB is close to or around 300 billion keys per second. I'm going after a 256 bit key with cpu now.

fxsniper

member

Activity: 406

Merit: 47

Quote from: WanderingPhilospher on April 26, 2021, 08:28:25 PM

Range is too high and too wide for a reason...Billions of dollars in wallets/private keys. If it was too narrow of a range, no one would use Bitcoin.

It's all relative to your GPU/CPU power. You can't have one GTX 1050 and expect to find keys in 120 bit range over night, or over a decade. But if you have many GPUs, then you have a chance to find keys after time, depending on your GPU power.

challenge is too big for 256 bit it is can not possible find

challenge what maximum we can do on records, will be stuck with 120 bit forever same 256 bit or not?

if can do 120 may be can use up to 140 bit or some 2-3 level, do far from 160 bit

I think all random method is stuck on too high range, next method may be concern about calculate or mix

last records is 256 card x Tesla V100 32GB right, How many power for GPU need to solve

I try kangaroo cpu only is difference with GPU over 12-15 time, kangaroo require to use with GPU

WanderingPhilospher

full member

Activity: 1162

Merit: 237

Shooters Shoot...

Quote from: fxsniper on April 26, 2021, 08:23:46 PM

Quote from: jacky19790729 on April 26, 2021, 11:37:54 AM

you must use %s , not %d

Improve and optimize the project "Kangaroo , VanitySearch " , maybe requires several years of learn VC++

Thank you

Yes, I need to learn C++ more

kangaroo is powerful random search

I think kangaroo need to modify to haves new feature can find key

or some tools like GPS find zone ok key and use kangaroo scan, now range is too high too wide range

other way is new one algorithm power more than kangaroo

Range is too high and too wide for a reason...Billions of dollars in wallets/private keys. If it was too narrow of a range, no one would use Bitcoin.

It's all relative to your GPU/CPU power. You can't have one GTX 1050 and expect to find keys in 120 bit range over night, or over a decade. But if you have many GPUs, then you have a chance to find keys after time, depending on your GPU power.

fxsniper

member

Activity: 406

Merit: 47

Quote from: jacky19790729 on April 26, 2021, 11:37:54 AM

you must use %s , not %d

Improve and optimize the project "Kangaroo , VanitySearch " , maybe requires several years of learn VC++

Thank you

Yes, I need to learn C++ more

kangaroo is powerful random search

I think kangaroo need to modify to haves new feature can find key

or some tools like GPS find zone ok key and use kangaroo scan, now range is too high too wide range

other way is new one algorithm power more than kangaroo

WanderingPhilospher

full member

Activity: 1162

Merit: 237

Shooters Shoot...

Quote from: unclevito on April 26, 2021, 06:20:42 PM

Quote from: WanderingPhilospher on April 26, 2021, 05:19:13 PM

Quote

One problem is that when it goes to save I get an illegal memory operation and it shuts down. The few times it did save the save work was 1.5 GB (a whole lot of points) and after the 2nd or 3rd save it would get the illegal memory operation so i just have the save work for 3 days. i don't want to do the entire keyspace on these larger puzzles because the electric could go off or some other issue and have to start over, rather just do a smaller slice. The smaller keys did not do that because it found them before 3o seconds and 50 is not that big of a keyspace as 120

You shouldn't get a memory code on save. Here's what I would do. Run the precompiled version and see if it does the same thing; see if you get memory error or the MK/s rate is the same.

It's always better to do the whole slice, not smaller ones because you will actually wind up doing more work/longer run time breaking it up into smaller chunks AND the biggest issue is, you really won't know if the key is in a smaller slice unless you let it run for 6+ times the expected operations and even then, you still may not know.

What is the link to the compiled version. I will try it, must be the 256 bit version for comparison, that is the one I am using not the 128 version.

https://github.com/JeanLucPons/Kangaroo/releases/tag/2.2

dextronomous

full member

Activity: 431

Merit: 105

guys anyone got a way to add a find a text *string i am searching for the puzzle address while inside golang.

// Get compressed and uncompressed addresses
      caddr, _ := btcutil.NewAddressPubKey(public.SerializeCompressed(), &chaincfg.MainNetParams)
      uaddr, _ := btcutil.NewAddressPubKey(public.SerializeUncompressed(), &chaincfg.MainNetParams)

      // Print keys
      fmt.Printf("%x %34s %34s\n", padded, uaddr.EncodeAddress(), caddr.EncodeAddress())
      fmt.Printf( "count:%v\n", count );
   }

somewhere in the area in the text above. thanks so much in advance

the result of what i am doing is outputting this on my screen
000000000000000000000000000000000000000000000000c249fdd32778046b 1HCeANYVqecNUkQGVnczXdjq2PjgtBdvDo 1JxXytq4KSJFgMZit8gZcRiiUZCYiQdoNR
count:14000000000000001131

but then like if found, We Found It! yippie something like text.

unclevito

jr. member

Activity: 76

Merit: 4

Quote from: WanderingPhilospher on April 26, 2021, 05:19:13 PM

Quote

One problem is that when it goes to save I get an illegal memory operation and it shuts down. The few times it did save the save work was 1.5 GB (a whole lot of points) and after the 2nd or 3rd save it would get the illegal memory operation so i just have the save work for 3 days. i don't want to do the entire keyspace on these larger puzzles because the electric could go off or some other issue and have to start over, rather just do a smaller slice. The smaller keys did not do that because it found them before 3o seconds and 50 is not that big of a keyspace as 120

You shouldn't get a memory code on save. Here's what I would do. Run the precompiled version and see if it does the same thing; see if you get memory error or the MK/s rate is the same.

It's always better to do the whole slice, not smaller ones because you will actually wind up doing more work/longer run time breaking it up into smaller chunks AND the biggest issue is, you really won't know if the key is in a smaller slice unless you let it run for 6+ times the expected operations and even then, you still may not know.

What is the link to the compiled version. I will try it, must be the 256 bit version for comparison, that is the one I am using not the 128 version.

WanderingPhilospher

full member

Activity: 1162

Merit: 237

Shooters Shoot...

Quote

One problem is that when it goes to save I get an illegal memory operation and it shuts down. The few times it did save the save work was 1.5 GB (a whole lot of points) and after the 2nd or 3rd save it would get the illegal memory operation so i just have the save work for 3 days. i don't want to do the entire keyspace on these larger puzzles because the electric could go off or some other issue and have to start over, rather just do a smaller slice. The smaller keys did not do that because it found them before 3o seconds and 50 is not that big of a keyspace as 120

You shouldn't get a memory code on save. Here's what I would do. Run the precompiled version and see if it does the same thing; see if you get memory error or the MK/s rate is the same.

It's always better to do the whole slice, not smaller ones because you will actually wind up doing more work/longer run time breaking it up into smaller chunks AND the biggest issue is, you really won't know if the key is in a smaller slice unless you let it run for 6+ times the expected operations and even then, you still may not know.

WanderingPhilospher

full member

Activity: 1162

Merit: 237

Shooters Shoot...

Quote from: unclevito on April 26, 2021, 05:02:00 PM

Yes I compiled it a few days ago and noticed it was going extremely fast so I tried it on a few puzzles that had been solved 30 35 40 and it found them on seconds except for 50 which took about 20 minutes so it seems to be working (used the entire keyspace for those). I was scanning the entire keyspace. this time I am trying 120 and I was just scanning 65% to 75% which was estimated at 15 hrs, now 50% to 65% which shows an estimate of 18 hrs. not sure if I should just scan the entire keyspace at once which has a 6 to 7 day estimate

I just don't think something is right, with MK/s, which will affect your solve time; meaning it will show a few days when it could be a few years.

For lower bits like 30, 35, 40, those will be quick, even on a CPU it will take less than a second. 50 shouldn't take that long, even with CPU.

What GPU or CPU are you using?

unclevito

jr. member

Activity: 76

Merit: 4

Yes I compiled it a few days ago and noticed it was going extremely fast so I tried it on a few puzzles that had been solved 30 35 40 and it found them on seconds except for 50 which took about 20 minutes so it seems to be working (used the entire keyspace for those). I was scanning the entire keyspace. this time I am trying 120 and I was just scanning 65% to 75% which was estimated at 15 hrs, now 50% to 65% which shows an estimate of 18 hrs. not sure if I should just scan the entire keyspace at once which has a 6 to 7 day estimate

One problem is that when it goes to save I get an illegal memory operation and it shuts down. The few times it did save the save work was 1.5 GB (a whole lot of points) and after the 2nd or 3rd save it would get the illegal memory operation so i just have the save work for 3 days. i don't want to do the entire keyspace on these larger puzzles because the electric could go off or some other issue and have to start over, rather just do a smaller slice. The smaller keys did not do that because it found them before 3o seconds and 50 is not that big of a keyspace as 120

Topic: Pollard's kangaroo ECDLP solver - page 59. (Read 58537 times)