Topic: BitCrack - A tool for brute-forcing private keys - page 69. (Read 77647 times)

3070 with 45 256 2096 - 1845Mkeys/s

Ive tried to make the same modifications to the VS project files and it wont compile for me.

Also i am not familiar enough with VS to begin with.

One side note, I had other GPUs on same rig when I rolled back driver. So I didn't get the error. I know it works, I had to roll back for helping with 64 bit pool effort, and then roll forward when using another program. 456 also works.

What speed did you see with 3070?

Source code available on github, to compile for Windows?

I have compiled cuBitCrack using CUDA 11.2, confirmed working on 2080, 2080ti, and 3070 without the multiplication issue preventing private keys from being found.

Can anyone help me create a windows binary?

https://github.com/yoyodapro/BitCrack/releases/tag/v11.2-alpha

Nvidia driver 452.06 doesn't support RTX 30 series, but 456.71 does support RTX 3090 and 3080. Windows 10 64-bit is listed as supported for both of these versions too.

I guess you should revert Nvidia drivers back to the latest version or whichever you were previously running with, now that we know that Bitcrack has a programming error.

The same happens with all 30series GPUs

It's currently possible to set the private key range by setting the keyspace for all of the addresses at once, not individually. Though that would make for an interesting side project.

You need to pass the --keyspace parameter with one of the following values:


Please note: Keys shorter than 64 characters will have zeroes padded at the beginning according to the documentation. So 1234 becomes 000001234. And keys longer than 64 characters will only use the rightmost 64 characters in the string.

Guys, please tell me with "xxBitCrack" whether it is possible to search for privkeys in one range from a list of several addresses in a file?
For this, what command with parameters do I need to register in the terminal?

It's not copyBigInt() itself that's problematic (it's a simple element-wise assignment)  but one of the arrays passed to it which is not aligned. CUDA wants all arrays aligned to 32-but boundaries and one of the arrays that eventually reaches copyBigInt() comes from "xp" and "x" pointer arguments of beginBatchAdd()...these are passed to SubModP() and the result is stored in an 8-element int array that's then passed to MulModP() and from there to copyBigInt().

At first it wasn't clear to me where this error was coming from because the problem disappeared in debug mode, so I could not use the debugger. That's right, if you pass -g -G switches to NVCC, you get a working but extremely slow bitcrack binary.

I tried draconian measures in a attempt to fix this like unrolling the loop, changing the array assignment to memcpy(), qualifying it with __restrict__ and __align__ keywords and I even changed it to a #define statement but the destination and source arrays just don't want to be accessed (since these arrays cannot even be used in the parent function, the problem stems deeper). More bafflingly, assigning a constant to an element in the dest array or making a variable that's initialized to an element from src works but this obviously breaks the elliptic curve stuff.

This is supposed to be performance-critical code so I did not attempt to change the static array to malloc.

---

For the uninitiated: this is where the bug is: https://github.com/brichard19/BitCrack/blob/master/cudaMath/secp256k1.cuh

CudaMath/secp256k1.cuh, everything in here are inline functions.

We arrive here from CudaKeySearchDevice via beginBatchAdd() and beginBatchAddWithDouble(). Both of these functions call MulModP for point multiplication. Methods like that need to copy to and from temporary arrays. Somehow the arrays being passed are not on an alignment boundary, and I'm honestly not sure what to do. (Of course, rewriting the whole secp256k1 module is also an option but really...? That's like opening a nut with a sledgehammer.)

What in/why is the copyBigInt() causing error?  Same error appears in multiple programs written prior to release of RTX 30xx cards.

Really appreciated it.. all the best.. finally, after a long time, we will be able to see a new update in CuBitCrack.

Guys I am busy debugging the misaligned address error using someone's GPU. I will post if I find the solution.

EDIT: Everyone please check if this error still appears if you pass the -u command line argument, it only seems to be failing for compressed points.

EDIT 2: pin-pointed the error to beginBatchAdd() and beginBatchAddWithDouble() in secp256k1 CUDA

EDIT 3: copyBigInt() in secp256k1.cu is the culprit

That video is too much, my ribs hurt   
Thank you for the entertainment

You need to run the program from the command prompt (Navigate using Dos commands eg: https://www.computerhope.com/overview.htm)
Like this screenshot
http://user-images.githubusercontent.com/45288590/48950290-f3aad180-ef3a-11e8-99ea-5e7024b2667e.png

Readme has usage here :
https://github.com/brichard19/BitCrack

Good Luck 

Sometimes I also get this error when I compile on servers with CUDA 10.0 and Nvidia drivers 450.57 preinstalled, but I cannot reliably reproduce it. I could not test this on 10.1 and later because of incompatibilities with the OS image on the server I ran these tests on. Most of the times it succeeds, but on some other identical servers it fails.

If anyone's willing to give me SSH access to a server that has an Nvidia GPU on it (renting one from VSYS.host will cost me at least $250/month which is out of my budget), ideally with CUDA and cuda-gdb already working and set up, I could try to see where the error is made. Misaligned address is an application error which is thrown by user code when it does something wrong.

this is a specific problem i decided to take a screenshot i just installed cuda 10.1  https://drive.google.com/file/d/1gwEMuUztlYyRe3tz2SFVr3QfRfguuT-Y/view?usp=sharing

Roll back your graphics driver to 452.06; only way it will work. Or the only way that I and others have got it to work

reason of this error, please ??

[2021-01-06.01:35:22] [Info] Compression: compressed
[2021-01-06.01:35:22] [Info] Starting at: 0000000000000000000000000000000000000000000000000000000058CE14F9
[2021-01-06.01:35:22] [Info] Ending at:   00000000000000000000000000000000000000000000000FFFFFFFFFFFFFFFFF
[2021-01-06.01:35:22] [Info] Counting by: 0000000000000000000000000000000000000000000000000000000100000000
[2021-01-06.01:35:22] [Info] Initializing GeForce RTX 3060 Ti
[2021-01-06.01:35:22] [Info] Generating 33,554,432 starting points (1280.0MB)
[2021-01-06.01:35:27] [Info] 10.0%
[2021-01-06.01:35:28] [Info] 20.0%
[2021-01-06.01:35:28] [Info] 30.0%
[2021-01-06.01:35:28] [Info] 40.0%
[2021-01-06.01:35:28] [Info] 50.0%
[2021-01-06.01:35:29] [Info] 60.0%
[2021-01-06.01:35:29] [Info] 70.0%
[2021-01-06.01:35:29] [Info] 80.0%
[2021-01-06.01:35:29] [Info] 90.0%
[2021-01-06.01:35:29] [Info] 100.0%
[2021-01-06.01:35:29] [Info] Done
[2021-01-06.01:35:29] [Info] Loading addresses from 'C:/Users/Desktop/1.txt'
[2021-01-06.01:35:29] [Info] 105 addresses loaded (0.0MB)
[2021-01-06.01:35:29] [Info] Allocating bloom filter (0.0MB)
[2021-01-06.01:35:29] [Info] Error: misaligned address

I didn't ask you to install Linux (in fact deploying bitcrack, specifically CUDA, on Linux is harder than on Windows) I'm only asking for your error messages that you received from bitcrack.


It is not possible to run bitcrack in a virtual machine (without doing GPU passthrough which is an advanced and expert-level procedure) or in emulation software such as Cygwin because the GPU cannot be accessed from there.

cygwin?
or virtualbox

i use windows 10 i don't have any usb memory to install linux.