Topic: BitCrack - A tool for brute-forcing private keys - page 14. (Read 74450 times)

For me the prize is the main, considering that i am investing money (actually around 400$ for hardware) and electricity costs (in Italy 0,55$ kilowatt/hour) in the challenge so a pool is not a good investment, the prize is too small.
Of course the glory of finding the key is nice and funny too but secondary.
I like to pick up randomly everyday the keyspaces to analyze, it's a way to see how the random choice progressively covers the entire puzzle 66 keyspace. For me it's very interesting.
Dividing the keyspace in 74 million parts it's like meeting a person in a country with 74 millions people.
With 7,4 million parts "only" it will be like meeting the same person in a big city (i will need a 2080ti to try it, actually price is around 400$).

Keyhunt-cuda brute forces ethereum too. I never tried it.

Tell please analog soft from Ethereum

Does anyone have a working copy of BitCrack2 from secp8x32 ? Please upload it somewhere so I could download a copy of the latest git tree. Thank you in advance

So if you had a supercomputer you would basically create a pool of your own resources, but no-one has that much hash - So why not join a pool and build a supercomputer together?

To me, cracking the puzzle is not really about the prize but I find it exciting to see how long it takes and what methods people use to crack it. like picking a lock. Plenty of people pick locks as a hobby

Just now i'm running a new keyspace check!  
And i keep notice of the checked keyspace.
I do it 5-6 times daily. It takes an hour each.

D:\BitCrack>cubitcrack -o out.txt --keyspace 20000000000000000:3ffffffffffffffff 13zb1hQbWVsc2S7ZTZnP2G4undNNpdh5so -b 32 -t 128 -p 5344 --share 36110528/74000000
[2022-11-04.07:14:56] [Info] Share 36110528 of 74000000
[2022-11-04.07:14:56] [Info] Compression: compressed
[2022-11-04.07:14:56] [Info] Starting at: 000000000000000000000000000000000000000000000002F9D886E6C2425E70
[2022-11-04.07:14:56] [Info] Ending at:   000000000000000000000000000000000000000000000002F9D8875AD6CA2BFF
[2022-11-04.07:14:56] [Info] Counting by: 0000000000000000000000000000000000000000000000000000000000000001
[2022-11-04.07:14:56] [Info] Initializing NVIDIA GeForce GTX 1060 3GB
[2022-11-04.07:14:56] [Info] Generating 21,889,024 starting points (835.0MB)
[2022-11-04.07:15:04] [Info] 10.0%
[2022-11-04.07:15:06] [Info] 20.0%
[2022-11-04.07:15:08] [Info] 30.0%
[2022-11-04.07:15:08] [Info] 40.0%
[2022-11-04.07:15:08] [Info] 50.0%
[2022-11-04.07:15:08] [Info] 60.0%
[2022-11-04.07:15:09] [Info] 70.0%
[2022-11-04.07:15:09] [Info] 80.0%
[2022-11-04.07:15:09] [Info] 90.0%
[2022-11-04.07:15:10] [Info] 100.0%
[2022-11-04.07:15:10] [Info] Done
NVIDIA GeForce G 2633 / 3071MB | 1 target 130.57 MKey/s (1,926,234,112 total) [00:00:13]

Before the year's end i'm planning to buy a 2080ti to brute force the keyranges like now (i should reduce the keyspace in 7,4 millions parts, now is 74 millions parts) and use the actual 1060 for the keyhunt-cuda random check, only for the unchecked keyspaces between the already checked keyspaces.

Cubitcrack uses brute force to analyze key by key the entire keyspace that is submitted to it.
Keyhunt-cuda allows you to search for the key within a predetermined keyspace in random mode.

what is the main difference between cubitcrack and keyhunt-cuda ?

I think the best strategy is to use 2 computers (or splitting the power of the supercomputer).
The first one uses cubitcrack and you select randomly the keyspaces to check daily. Small keyspace portions to cover progressively the entire puzzle 66 range and without concentrating on few big portions.
The second one uses keyhunt-cuda in random mode to check for a fixed period of time (e.g. 12 hours) the unchecked keyspaces between the already checked keyspaces.

About commands you can see them in my previous posts.

Let's say you would have something like a 'supercomputer' with enormous power, what bitcrack command would you use to run for trying to crack the puzzle(s) ? Would you use cubitcrack or even bitcrack2 ?

My second question is: Are there any available tools like BitCrack that will make use of Tensor for much higher calculation rates than CUDA, does it make sense at all for Tensor use in our applications? As far as I know, Tensor cores can compute a lot faster than the CUDA cores. CUDA cores perform one operation per clock cycle, whereas tensor cores can perform multiple operations per clock cycle. On the other hand, CUDA cores produce very accurate results. So it Tensor only good for AI learning and NN but not for brute-force cracking for our application?

Better to try luck alone, perhaps sharing the ranges already tested. The prize is too small to split into a group. I believe that a random approach in choosing only the intervals is the best solution because it allows you to test small portions of the entire range and make a good competition to groups organized in a pool and which are usually concentrated on large and not very distributed portions..
Let's see who will find the key first!

There are just so many ranges on Puzzle #66 that its near impossible to solo.

Stats from ttdsales pool below in my sig.

Total Ranges Checked: 92,272 of 33,554,432
Total Private Keys Checked: 101.45414 Quadrillion
Percentage completed: 0.274992%

the ttd pool is using a customised vanitygen for cracking, source is available in the files on the site. seems to work quicker than Bitcrack. Worth trying out even if not with the pool, I found big improvements on the 2080 Ti's and 3070's not as much on the 3090's but still faster.

http://www.ttdsales.com/66bit/index.php

@ETFbitcoin: thank you. Wasn't aware about this compare feature, will dig into it later. Thanks also for your effort in highlighting the differences.

If anyone still has the original .zip file of BitCrack2 of secp8x32 please upload it somewhere and let me know where I can download it. You can also reach me via DM if needed.

Not down, but either deleted or set to private by owner of the repository.


You could try your luck using Wine. Last time i ran VanitySearch.exe using Wine since i didn't want to compile from source code and it run without problem.


Using GitHub comparison feature[1] and read a bit of commit message/README.md file, here are few difference i found
1. brichard19's version support CUDA 10.1 while djarumlights's version support CUDA 10.2
2. modification on djarumlights's version only meant for CUDA usage, although it's not tested extensively.
3. djarumlights's version add feature random with following description.


[1] https://github.com/brichard19/BitCrack/compare/master...djarumlights:BitCrack:master

What happened to BitCrack2 that supported random mode

https://github.com/secp8x32/BitCrack2

Link is down ? anyone can point me to another download source ?
I found another git source of BitCrack2 from djmuratb but it seems that there is no Linux compatibility?

And what exactly is the difference between original BitCrack Version from brichard19 and this BitCrack version from djarumlights ?

I think you should try the same parameters of the youtube videos: -b 128 -t 512 -p 1024
You should increase the Mkey/s.
Personally with my 1060 3GB (120 Watt) the best parameters are -b 32 -t 128 -p 5344
I reach 125/130 Mkey/s

These are the puzzle 66 keyspaces already checked (one hour random lottery tickets, 500 billions keys each, only few of them are smaller because i checked them before buying the 1060):

3F926FA3FC924E170   3F926FAB3DDACAEFF
3ED3F6ACA58CB83E0 3ED3F6B3E6D53516F
3E8F8E769B265FC70   3E8F8E7DDC6EDC9FF
3D7C470AF46D1A840 3D7C471235B5975CF
3C779266255CCA1D0 3C77926D66A546F5F
3C5BDD67E5793C2A0 3C5BDD6F26C1B902F
3BE08910DDC9983E0 3BE089181F121516F
3BCFDADE1E8E05180 3BCFDAE55FD681F0F
3BC7FD25F6B0FE080   3BC7FD2D37F97AE0F
3B9816FCB89BF86E0   3B981703F9E47546F
3B0499A2C809C09E0   3B0499AA09523D76F
3A7553158FC497D08   3A755315DA4614500
39431292DDCB40EC0 3943129A1F13BDC4F
39121EBAED3569060   39121EC22E7DE5DEF
38CBE30633E2319F0   38CBE30D752AAE77F
38C12420AA9779530   38C12427EBDFF62BF
388D7C0DCBA968930 388D7C150CF1E56BF
38141628F72E6DB60   381416303876EA8EF
37B30F35556E17D40   37B30F3C96B694ACF
369BEAE5484B80730   369BEAEC8993FD4BF
35DE6DAC378A08550 35DE6DB378D2852DF
3591A0DF6ABDC2CF0 3591A0E6AC063FA7F
358E38DAB00FA86D0 358E38E1F1582545F
34BB97BB72E3B07B0 34BB97C2B42C2D53F
33801AF3842571250   33801AFAC56DEDFDF
335BDD67E58916760 335BDD6F26D1934EF
32419900565E240D0   3241990797A6A0E5F
321F0AFCCA5BC8A10   321F0B040BA44579F
32026C46744A480F0   32026C4DB592C4E7F
3158E4602CA7CB207   3158E4607729479FF
3152AA13D28903980   3152AA1B13D18070F
3010543F566379F00   3010544697ABF6C8F
2FDAF3AFCF614E110   2FDAF3B710A9CAE9F
2F97C61AEA9140480   2F97C6222BD9BD20F
2F66D24710E82C359   2F66D2475B69A8B51
2F085B8C68F7EED40   2F085B93AA406BACF
2E9E0649EEA3B9440   2E9E06512FEC361CF
2E63D61AD2FF331A0   2E63D6221447AFF2F
2E10B72ADDC0CC811 2E10B72B284249009
2CE5D68FBACFAD341 2CE5D690055129B39
2C1203CAB547C3796 2C1203CAFFC93FF8E
2C0487AB1A9EFB080   2C0487B25BE777E0F
2B0E77CC12054DC00 2B0E77D3534DCA98F
2B057A6648D9C6210 2B057A6D8A2242F9F
2A7376979E7882CB0   2A73769EDFC0FFA3F
298F588851734C0C0   298F588F92BBC8E4F
2955C12F7DDCDD4B0 2955C136BF255A23F
288EFEA36B4A1E340   288EFEAAAC929B0CF
2879C8C5915A6CF60   2879C8CCD2A2E9CEF
284D89926FD6A8630   284D8999B11F253BF
27746952CCB6D3BC0 2774695A0DFF5094F
26A82C607136EE600   26A82C67B27F6B38F
26981FFFD0CC0B580   2698200712148830F
26270CB2CFE94F470   26270CBA1131CC1FF
2611D6D7C835A9FE9 2611D6D812B7267E1
25C9EB9F6B54E4380   25C9EBA6AC9D6110F
24D26B29B2045E520   24D26B30F34CDB2AF
24D0BBA9A6B947FBF 24D0BBA9F13AC47B7
2495BCB533282D3E0 2495BCBC7470AA16F
247EFB4B3458D250B   247EFB4B7EDA4ED03
23EB35FEB9B965400   23EB3605FB01E218F
239DA36E8CA682EEF   239DA36ED727FF6E7
230E1DE72BE8745C0   230E1DEE6D30F134F
229C9EB6560D19440   229C9EBD9755961CF
21D4E0733E7EB94B0   21D4E07A7FC73623F
213FA19B123A46984   213FA19B5CBBC317C
210717EF6CB7B4590   210717F6AE003131F
20DEA3A1C2CF93A10 20DEA3A904181079F

 

Can you help me to understand why on cubitcrack an rtx 2080ti has quite the same performance of an rtx 3080ti?  
If you see these 2 videos you see that the rtx 2080ti reaches 1480Mkeys/s and the rtx 3080ti 1688Mkeys/s

https://www.youtube.com/watch?v=gCgdKkXcn9g
https://www.youtube.com/watch?v=UNwMctHWkPI

How is it possible? there should be a difference of 10x between these 2 video cards..

4352 vs 10240 cuda cores
18,600 vs 28,300 million transistors
270 vs 350 Watts

https://technical.city/it/video/GeForce-RTX-2080-Ti-vs-GeForce-RTX-3080-Ti

The price of these 2 cards is very different (used 500$ vs 1000$).

It could be a surprise if cards like QUADRO RTX 8000 48gb GDDR 6 (used 4000$) or a FirePro Graphics W7000 (used 80$) could give better (or worse results).

What really makes the difference to have the best performance on cubitcrack (or keyhunt-cuda)? the cuda cores only? the ram?

Thanks for reply! 

I'd like to understand how to increase the Mkeys/s on cubitcrack/keyhunt-cuda.

I have a ASUSTeK CROSSHAIR VI HERO (AM4) motherboard with Multi-GPU SLI / CFX support :
2 x PCIe 3.0 x16 SafeSlots (CPU)

It means that i can install 2 GPUs.

Now I have one GTX 1060 3GB.
Can i add on the same motherboard another different video card like a 2080 or 3070 or 4090? 
Or do i need to install one similar like another 1060 or 1080? 

I read that 2 gpus can work in SLI mode, in cooperation mode. 
Or is it better to make them work independently with these softwares? 

The PSU is 850W so it should be enough.

What do you think?

Not a lot, but at least you could avoid wasting CPU cycles on M$hit.

Now I have a better computer to make tests (desktop Ryzen 7 1700 8 cores 16 threads + Geforce GTX 1060 3GB).
It's very silent and it doesn't heat up much (the case is a Cooler Master with 850W PSU).  

Here the stats:
Cubitcrack : 120 Mkeys/s (with peaks of 123/125 Mkeys/s)
keyspace of 500 billions keys is completed in 1,08 hour.

D:\BitCrack>cubitcrack -o out.txt --keyspace 20000000000000000:3ffffffffffffffff 13zb1hQbWVsc2S7ZTZnP2G4undNNpdh5so -b 320 -t 256 -p 256 --share 25486986/74000000
[2022-10-13.14:35:01] [Info] Share 25486986 of 74000000
[2022-10-13.14:35:01] [Info] Compression: compressed
[2022-10-13.14:35:01] [Info] Starting at: 000000000000000000000000000000000000000000000002B057A6648D9C6210
[2022-10-13.14:35:01] [Info] Ending at:   000000000000000000000000000000000000000000000002B057A6D8A2242F9F
[2022-10-13.14:35:01] [Info] Counting by: 0000000000000000000000000000000000000000000000000000000000000001
[2022-10-13.14:35:01] [Info] Initializing NVIDIA GeForce GTX 1060 3GB
[2022-10-13.14:35:01] [Info] Generating 20,971,520 starting points (800.0MB)
[2022-10-13.14:35:17] [Info] 10.0%
[2022-10-13.14:35:20] [Info] 20.0%
[2022-10-13.14:35:21] [Info] 30.0%
[2022-10-13.14:35:22] [Info] 40.0%
[2022-10-13.14:35:22] [Info] 50.0%
[2022-10-13.14:35:23] [Info] 60.0%
[2022-10-13.14:35:24] [Info] 70.0%
[2022-10-13.14:35:24] [Info] 80.0%
[2022-10-13.14:35:25] [Info] 90.0%
[2022-10-13.14:35:26] [Info] 100.0%
[2022-10-13.14:35:26] [Info] Done
NVIDIA GeForce G 2530 / 3071MB | 1 target 120.02 MKey/s (498,367,201,280 total) [01:07:37][2022-10-13.15:43:07] [Info] Reached end of keyspace

Keyhunt-cuda : 110 Mkeys/s (with peaks of 113 Mkeys/s)
keyspace of 500 billions keys is completed in 1,16 hour.

D:\Keyhunt>KeyHunt-Cuda.exe -g -m addresses --coin BTC --range 210717EF6CB7B4590:210717F6AE003131F -i puzzle66.bin

KeyHunt-Cuda v1.08

COMP MODE    : COMPRESSED
COIN TYPE    : BITCOIN
SEARCH MODE  : Multi Address
DEVICE       : GPU
CPU THREAD   : 0
GPU IDS      : 0
GPU GRIDSIZE : -1x128 (Auto grid size)
SSE          : YES
RKEY         : 0 Mkeys
MAX FOUND    : 65536
BTC HASH160s : puzzle66.bin
OUTPUT FILE  : Found.txt

Loaded       : 1 Bitcoin addresses

Bloom at 00000277594F7BE0
  Version    : 2.1
  Entries    : 2
  Error      : 0.0000010000
  Bits       : 57
  Bits/Elem  : 28.755175
  Bytes      : 8 (0 MB)
  Hash funcs : 20

Start Time   : Thu Oct 13 10:20:11 2022
Global start : 210717EF6CB7B4590 (66 bit)
Global end   : 210717F6AE003131F (66 bit)
Global range : 741487CD8F (39 bit)

GPU          : GPU #0 NVIDIA GeForce GTX 1060 3GB (9x128 cores) Grid(72x128)

[01:15:51] [CPU+GPU: 110.09 Mk/s] [GPU: 110.09 Mk/s] [C: 100.512239 %] [R: 0] [T: 501,114,470,400 (39 bit)] [F: 0]

BYE

Thanks to the faster gpu i divided the puzzle 66 keyspace in 74 millions parts, each part can be tested in a little more than an hour). Below the keyspaces already checked (so it's not necessary for you to check it twice):

3D7C470AF46D1A840:3D7C471235B5975CF

3A7553158FC497D08:3A755315DA4614500

3158E4602CA7CB207:3158E4607729479FF

2F66D24710E82C359:2F66D2475B69A8B51

2E10B72ADDC0CC811:2E10B72B284249009

2CE5D68FBACFAD341:2CE5D690055129B39

2C1203CAB547C3796:2C1203CAFFC93FF8E

2B057A6648D9C6210:2B057A6D8A2242F9F

2611D6D7C835A9FE9:2611D6D812B7267E1

24D0BBA9A6B947FBF:24D0BBA9F13AC47B7

247EFB4B3458D250B:247EFB4B7EDA4ED03

239DA36E8CA682EEF:239DA36ED727FF6E7

213FA19B123A46984:213FA19B5CBBC317C

210717EF6CB7B4590:210717F6AE003131F

Now i do the tests in the windows 10 terminal.
Do you think that with linux (like Ubuntu) i could increase a lot the Mkeys/s performance?

It depends on the computational resources of the participants, with a large participation maybe in a couple of years we will solve it!