Encryption Projects as SU group

cayoenrique · Aug 17, 2023

I been Hijacking others thread. And I wan to stop that bad habit. Here lets open this thread to discuss any subject we have. As you see objective of this tread is OPEN. Feel free to share your thought.

cayoenrique · Aug 17, 2023

I know we lost many post. And I do not recall the status of a few.

dvlajkovic and I where mentioning CUDABiss well go next after the 1rts one. fiji's CW Finder .

Let me start by asking about ( CW Finder C-8 Tools CW / C-8 Update List )
I want to know how fast can you get it working.

Lets use a standard measurement MKPS = Mega Keys search Per Seconds = 1,000,000/seg = 10^6/s
Please describe as best as you can your PC system at least, CPU + Memory.
And if you have an SDD please tell us if it improves.

Hope some join in.

cayoenrique · Aug 17, 2023

Here this was what I did used on the past

Code:

Form factor: laptop, Desktop
Operating System : Win7, Win10
Ram Memory size: 4/8 GB
MAin Storage: HDD, SDD 
CPU model Speed and number of Cores: Intel 4 Core @ 2.00GHz
GPU model:
KPS on CPU:
KPS on GPU:

**KPS: Average Keys/seconds. Use MKS for mega keys per second (10^6).
For example:
987,543 => 0.987 MKPS
4,467,961 => 4.468 MKPS
This way it is easyly to compare.

For example my everyday PC is

Code:

Form factor: laptop
Operating System : Debian BusterDog
Ram Memory size: 8 GB
MAin Storage: HDD 
CPU model Speed and number of Cores: Intel 4 Core @ 2.00GHz
GPU model: Radeon HD 6730M 
KPS on CPU: 20.029 MKPS in aycwabtu
KPS on GPU:  N/A

dvlajkovic · Aug 18, 2023

CW Finder 1.0
The search using BF (without cw list)
Options > Set Process Priority > Highest
Options > Set Search Thread Priority > Highest

I ran the search for exactly 1 minute and got this result:

This means for 1 minute

achieved speed is 4,641,296 CW/s (4.641 MKPS) in this app
this app checked 278,477,776 keys
and got from 00 00 00 00 00 00 to 00 00 10 B6 90 E0.
The remaining time to perform full search is 01 years 11 months 6 days 22 hours 01 minute and 51 seconds.

Presence of encrypted .ts file on SSD means nothing to this kind of search.

My PC is an older rig running on Win10 64bit, nothing exceptional nowadays:

CPU AMD Ryzen 9 3950x
RAM KINGSTON HyperX Fury RGB 32 GB (2x16 GB) DDR4 3466 MHz CL16
SSD PCIe 4.0 M.2 NVMe Gen4 Aorus Gigabyte

cayoenrique · Aug 18, 2023

Before I start. This is to every one. My intention is to share what I know with all of you. I am not trying to show up or even less humiliate anyone. In this tread feel free to ask. there is no wrong question. Just ask.

@dvlajkovic

(4.641 MKPS) , I believe that is the original. Fiji build a new one. I do not know if it is faster.

Now this is what I did in lesson 9 of the encryption learning. I can do around 20 MKPS

Code:

Form factor: laptop
Operating System : Debian BusterDog
Ram Memory size: 8 GB
MAin Storage: HDD 
CPU model Speed and number of Cores: Intel 4 Core @ 2.00GHz
GPU model: Radeon HD 6730M 
KPS on CPU: 20.029 MKPS in aycwabtu
KPS on GPU:  N/A[/ccode]

The original aycwabtu can not do Search on CW_list_Attack.  You can find it
[code]https://github.com/aycwabtu/aycwabtu

And here is the Open Source program we build on top of aycwabtu.

The 1rst link will allow you to download everything.

Code:

dEvPpb6TYz 28.11 MB
https://workupload.com/archive/dEvPpb6TYz

You can download separately, Source, bin & sample attack list.

Code:

aycwabtu_malvanos_4.11_sources.zip (79.73 KB)
https://workupload.com/file/FzhgFTZTUFu

Code:

aycwabtu_malvanos_4.11_bin.zip (5.38 MB)
https://workupload.com/file/gNLrMgGn5qS

Code:

CW_list_Attack.zip (22.65 MB)
https://workupload.com/file/szBSBr5wS9c

To build you can use codeblock IDE using aycwabtu-1.0-win.cbp

Or manually doing make

To execute you go to bin folder and place the extracted List attack called CW_list_Attack.txt or any that you have. Then use runme.bat. Or manually you can write

Code:

aycwabtu-1.0.exe aycwabtu_Testfile_PID_123_CW_7FFAE9A02486.ts CW_list_Attack.txt

And you will see:

Code:

E:\CWD\aycwabtu_malvanos_4.11_bin>aycwabtu-1.0.exe aycwabtu_Testfile_PID_123_CW_7FFAE9A02486.ts CW_list_Attack.txt
AYCWABTU CSA brute forcer V1.2a Aug 18 2023
CPU only - all bit slice (bool sbox)
parallel bitslice batch size is 128
----------------------------------------
Using brute force txt file

fsize = 140001078     140001078
01230000 01 23 00 00 01 30
0123FFFF 01 23 FF FF FE 10
using TS file aycwabtu_Testfile_PID_123_CW_7FFAE9A02486.ts
searching for encrypted packets...
found three encrypted packets with pid 123 (0x7B) cw parity 0:
4A 83 70 2B 9C CD 2B F6  D1 4F 08 20 20 20 20 20  
02 A6 6A 2C FD EC 1D F3  20 BD 58 20 20 20 20 0D  
A6 B3 8A D1 DB B0 BB B6  81 35 35 8E 20 20 20 20  
start key is 01 23 00 [] 00 00 00 []
stop key is  01 23 FF [] FF FF FF []
/ 18.957 Mcw/s avg: 19.355 Mcw/s  01 23 04 [] B9 .. .. []
key candidate successfully decrypted three packets
KEY FOUND!!!    7F FA E9 [62]  A0 24 86 [4A]
writing result to file "keyfound"

E:\CWD\aycwabtu_malvanos_4.11_bin>

If you see errors. Make sure you have in folder

Code:

aycwabtu_Testfile_PID_123_CW_7FFAE9A02486.ts
CW_list_Attack.txt
libgomp-1.dll

cayoenrique · Aug 18, 2023

So now you have sources to play and do your thing.

Next is How we can improve with GPU.

@dvlajkovic

Can you show us what you do with CUDABISS? So we have it as reference.

@all
I do not know Cuda nor I have CUDABISS sources. But I guess we can try to do our best. Who want to join in?

SatEze · Aug 18, 2023

This is a very informative and interesting topic.

cayoenrique said:
Before I start. This is to every one. My intention is to share what I know with all of you. I am not trying to show up or even less humiliate anyone. In this tread feel free to ask. there is no wrong question. Just ask.

@dvlajkovic

(4.641 MKPS) , I believe that is the original. Fiji build a new one. I do not know if it is faster.

Now this is what I did in lesson 9 of the encryption learning. I can do around 20 MKPS

Code:

Form factor: laptop Operating System : Debian BusterDog Ram Memory size: 8 GB MAin Storage: HDD CPU model Speed and number of Cores: Intel 4 Core @ 2.00GHz GPU model: Radeon HD 6730M KPS on CPU: 20.029 MKPS in aycwabtu KPS on GPU: N/A[/ccode] The original aycwabtu can not do Search on CW_list_Attack. You can find it [code]https://github.com/aycwabtu/aycwabtu

And here is the Open Source program we build on top of aycwabtu.

The 1rst link will allow you to download everything.

Code:

dEvPpb6TYz 28.11 MB https://workupload.com/archive/dEvPpb6TYz

You can download separately, Source, bin & sample attack list.

Code:

aycwabtu_malvanos_4.11_sources.zip (79.73 KB) https://workupload.com/file/FzhgFTZTUFu

Code:

aycwabtu_malvanos_4.11_bin.zip (5.38 MB) https://workupload.com/file/gNLrMgGn5qS

Code:

CW_list_Attack.zip (22.65 MB) https://workupload.com/file/szBSBr5wS9c

To build you can use codeblock IDE using aycwabtu-1.0-win.cbp

Or manually doing make

To execute you go to bin folder and place the extracted List attack called CW_list_Attack.txt or any that you have. Then use runme.bat. Or manually you can write

Code:

aycwabtu-1.0.exe aycwabtu_Testfile_PID_123_CW_7FFAE9A02486.ts CW_list_Attack.txt

And you will see:

Code:

E:\CWD\aycwabtu_malvanos_4.11_bin>aycwabtu-1.0.exe aycwabtu_Testfile_PID_123_CW_7FFAE9A02486.ts CW_list_Attack.txt AYCWABTU CSA brute forcer V1.2a Aug 18 2023 CPU only - all bit slice (bool sbox) parallel bitslice batch size is 128 ---------------------------------------- Using brute force txt file fsize = 140001078 140001078 01230000 01 23 00 00 01 30 0123FFFF 01 23 FF FF FE 10 using TS file aycwabtu_Testfile_PID_123_CW_7FFAE9A02486.ts searching for encrypted packets... found three encrypted packets with pid 123 (0x7B) cw parity 0: 4A 83 70 2B 9C CD 2B F6 D1 4F 08 20 20 20 20 20 02 A6 6A 2C FD EC 1D F3 20 BD 58 20 20 20 20 0D A6 B3 8A D1 DB B0 BB B6 81 35 35 8E 20 20 20 20 start key is 01 23 00 [] 00 00 00 [] stop key is 01 23 FF [] FF FF FF [] / 18.957 Mcw/s avg: 19.355 Mcw/s 01 23 04 [] B9 .. .. [] key candidate successfully decrypted three packets KEY FOUND!!! 7F FA E9 [62] A0 24 86 [4A] writing result to file "keyfound" E:\CWD\aycwabtu_malvanos_4.11_bin>

If you see errors. Make sure you have in folder

Code:

aycwabtu_Testfile_PID_123_CW_7FFAE9A02486.ts CW_list_Attack.txt libgomp-1.dll

This a very interesting and informative learning experience. What's the password to extract the files after downloading them?

Edit:
CPU: AMD Ryzen 7 5700X 8-Core Processor 3.40 GHz
GPU: MSI Gaming X Geforce RTX 3060, 12GB
RAM: 64.0 GB
No Overclocking yet

moonbase · Aug 18, 2023

CudaBISS.
Gigabyte RTX 3090 (Aorus Extreme) with 10496 cuda cores requires approximately 1 day to run a full range.

cayoenrique · Aug 18, 2023

How s7up1d of me. Sorry

My password will always be, present and pass files. Easy no. I put password as in my experience files last longer on web. I suspect as they are lock no one complains. But the password is generic always the same.

Code:

www.sat-universe.com

cayoenrique · Aug 18, 2023

@moonbase
Lets be clear. That table you made it. I guess it means you ran 4 instances at same time of Cudabiss? Is that what it means?

So you finally got 3378.38 mkps.

This means that you can do 1 day full search

Code:

2 ^ 48 = 281474976710656 different Keys in CSA
281474976710656	/	3378380000	=	83316.5531144087	sec
						60			1388.60921857348	min
						60			23.1434869762246	hrs

I guess this is what dvlajkovic said was possible! Interesting.

dvlajkovic · Aug 18, 2023

I have used CW Finder 1.0 to show speed of search and you've been using aycwabtu.
Please run CW Finder 1.0 on windows and post your score.

We are not looking to improve (speed up) search based on attack list.
Why? Because it works wonderful and there's no need to further develop this tool.
By "wondeful" I mean it finds cw in a few seconds.
Downside of search based on attack list: if correct cw is not among the millions of them stored in attack list, no cw will be found.
We already concluded that there is no way to generate and store a file that BIG to have all possible cws inside of an attack list, so please have this idea crossed out.
Instead, every user of search based on attack list keeps and updates attack lists for personal use. Some of these lists are published, but that's not even 1% of all possible cws

That is why we would appreciate to have a tool that does the real brute-force search (without any attack list) to cover to the full range fast.

By the way, I have already posted brute-force results of my rig using old cudabiss - I guess the "Resulting speed" info there never drew any attention.
Link 1
Link 2
Link 3

In short, dragon in my rig clocks 934,124 MKPS in just one instance.
You can imagine what a firestorm of MKPS it would be if I tamed the dragon to launch 4 instances at once

PS: I currently have a signed non-aggression pact with that beast and would prefer to keep it that way.

cayoenrique · Aug 18, 2023

dvlajkovic

Can you tell us what is this Dragon you mention. A model number will be nice so that I can read in the web about it.

Why I started with About attack list? It does not matter is it is a list or a consecutive search. Point is to show to the new guy a starting point speed with no GPU.
Regards you getting a new Cuda tool from me that can improve over what you have. I do not know Cuda. Nor have Nvidia GPU. Then what are the chances!!! I can not offer what I do not have.

What I offered was to create a group to learn and build tools. I was teach OpenCL by a old master. I can offer to teach that to others. Will it be faster than what you have? I do not know for sure. But I guess that if we try maybe we can prove that we can.

Bu the way I am still working on T2MI.

moonbase · Aug 18, 2023

cayoenrique said:
@moonbase
Lets be clear. That table you made it. I guess it means you ran 4 instances at same time of Cudabiss? Is that what it means?

So you finally got 3378.38 mkps.

Yes, I ran multiple instances of CudaBISS at the same time.

moonbase · Aug 18, 2023

dvlajkovic said:
By the way, I have already posted brute-force results of my rig using old cudabiss - I guess the "Resulting speed" info there never drew any attention.
Link 1
Link 2
Link 3

In short, dragon in my rig clocks 934,124 MKPS in just one instance.

From your links I read it as just over 934 Mkps from The Dragon.

I think there can be some confusion with the . or , character as the decimal place delimiter?
As I am in the UK I use the "dot" character as the decimal place delimiter. In Europe the "comma" is often used as the decimal place delimiter.
It might be useful if we all quote MKps speeds as rounded up/down to the nearest whole number?

dvlajkovic · Aug 18, 2023

cayoenrique said:
dvlajkovic

Can you tell us what is this Dragon you mention. A model number will be nice so that I can read in the web about it.

Amigo, please read the story below.
You'll find all the info to look it up on the Net.

moonbase said:
From your links I read it as just over 934 Mkps from The Dragon.

I think there can be some confusion with the . or , character as the decimal place delimiter?
As I am in the UK I use the "dot" character as the decimal place delimiter. In Europe the "comma" is often used as the decimal place delimiter.
It might be useful if we all quote MKps speeds as rounded up/down to the nearest whole number?

The rounded down number from my machine is 3378 MKps
Your dragon is breathing fire at about 27% of the speed of an RTX 3090 GPU in an i7-5930k machine

CPU has no use in cudabiss.
The Resulting speed info you've seen on those 3 links is copy/paste from cudabiss output. No delimiters added or removed.
Once you find more about this birdie, think again about eventual confusion.

I'm gonna speak today with my dragon.
Will try to explain with carefully chosen words what people are saying about his breath.
It's name is "Phantom", from a Golden Sample tribe, so I guess it's a male.
I've found him on Net sitting on the very top of its family tree, looking at me from the most left side, as a real predator.
Gotta admit, he was sooo unwilling to leave his family and fly over to some silly cave of mine.
That's why we signed a pact - will not disturb him unless absolutely necessary.
Because, who am I to challenge his arguments?

cayoenrique · Aug 18, 2023

"Golden Sample" is a series, maybe from RTX 3070 to 4090 + there with or not TI. It is not enough info.. It is not enough info.

moonbase · Aug 18, 2023

cayoenrique said:
"Golden Sample" is a series, maybe from RTX 3070 to 4090 + there with or not TI. It is not enough info.. It is not enough info.

My personal tests with cards from two series of GPU are listed below:
GTX 980Ti runs a single instance of CudaBISS at approximately 700 Mkps
GTX 1080Ti runs a single instance of CudaBISS at approximately 1060 MKps

moonbase · Aug 18, 2023

dvlajkovic said:
CPU has no use in cudabiss.
The Resulting speed info you've seen on those 3 links is copy/paste from cudabiss output. No delimiters added or removed.
Once you find more about this birdie, think again about eventual confusion.

My point about delimiters was to try to get any people who wished to share results to upload them in rounded whole numbers so they were comparable.
I did not say you added or removed any delimiters, all I did was to round your CudaBISS Mkps speed to a whole number, in this case, 934 Mkps.

You are incorrect in your statement that CPU has no use in CudaBISS. The CPU can determine the number of PCIe lanes available to the GPU.
For example, an i7-5820k CPU can only provide 28 PCIe lanes but an i7-5930k CPU can provide 40 PCIe lanes.

If two RTX 3080Ti Graphics cards are used in a motherboard with an i7-5930k CPU they can both run at full x16 speed.
The same two cards in the same motherboard fitted with an i7-5820k cannot both run at x16 speed, one of them can but the other one runs at x8 speed.
This has an impact on the CudaBISS Mkps speed for the card running at x8 speed, it is reduced.

How do I know this, because I tested an X99 motherboard fitted with two RTX 3080Ti cards with both these CPU's and measured the Mkps speeds.
The idea that CudaBISS speed in not dependent on CPU is incorrect. It may have been correct years ago with old CPU's and old graphics cards.
However, with modern graphics cards and 28 lane or 40 lane CPU's there is a contribution from the CPU.

cayoenrique · Aug 18, 2023

You guys can post all you reviews. I guess can help others know what they can get for the $.

My intention was to tell the newbie that CPU at best is on the 20 MKPS. But with the new graphics we can do

Drago GS rtx **** single Instance
Speed :934 mkps

Gigabyte RTX 3090 (Aorus Extreme) with 10496 cuda cores, 4 instances
Speed = 3378.38 mkps.

as explained. I did learned. I was not expecting a 23 hour Full search on 1 GPU. Just for your knowledge, there is a less use mechanism call FPGA that one of our friend use and produce less power consumption (Heat) but he still do full search in 24 hours.

moonbase · Aug 18, 2023

cayoenrique said:
...Just for your knowledge, there is a less use mechanism call FPGA that one of our friend use and produce less power consumption (Heat) but he still do full search in 24 hours.

There might be coders using FPGA and custom coded apps for key search, however, as far as I know these apps if they exist are not shared in the public domain.
My understanding was that a full search with FPGA could be done in 12 hours but I could be wrong on that?

Encryption Projects as SU group

cayoenrique

Member

cayoenrique

Member

cayoenrique

Member

dvlajkovic

Member

cayoenrique

Member

cayoenrique

Member

SatEze

moonbase

cayoenrique

Member

cayoenrique

Member

dvlajkovic

Member

cayoenrique

Member

moonbase

moonbase

dvlajkovic

Member

cayoenrique

Member

moonbase

moonbase

cayoenrique

Member

moonbase