Multiforcer for Windows 32 bits

[Sc00bz]

Remember, I generate table parts and merge them. That's what numtables specifies - how many N-sized tables are created, where they will later be merged.

This was one of my complaints about WinRTGen because it makes stupid people. You have file(s) that make a table and you have table(s) that make up a table set. The probably with this is people will plug in 1000 tables into a rainbow table calculator and think that's the success rate. WinRTGen is extremely bad since the "N° of tables" field is used for both number of tables and number of files.

Anyway now that I know that --numtables is all part of the same table I'm no longer one of the stupid people that spread bad info. (... stupid, ignorant whatever I like being stupid about ignorance )

Remember, I generate table parts and merge them. That's what numtables specifies - how many N-sized tables are created, where they will later be merged.

there's something i can't still catch on.. my intent is to recover one MD5 hash at a time, what difference if i create 1 table part or 1000 table parts?

The problem is that you can't hold a billion chains in GPU memory but you can hold a million chain in GPU memory and dump those to disk 1000 times. So when generating this you need to split the table into small enough chunks so you can work with.

One table part is easier to search than 1000 table parts which is why you merge the 1000 table parts into one.

[pellegrino_tarantino]

Ok, something is clearing in my mind.. so if i create a chain of 200000 hashes, i can choose in how many parts it will be splitted? If i chose a higher number of tables parts, each of them will have only a smaller dimension and will contain a part of the chain of 200000. Is it right?

edit: is there some way, except to change GPU, to avoid this in GRTGen-OpenCL

Code: Select all

ERROR: cannot allocate GPU memory. Try a smaller number of chains

i set --chainlenght to 200000 and --numchains 83885000 (this is the best value i can set without errors)

[Bitweasil]

I don't think you understand rainbow tables at all. I can't make sense of your questions, and I don't think it's from language issues.

Please go read and understand:

http://en.wikipedia.org/wiki/Rainbow_table

http://cryptohaze.com/wiki/index.php/Rainbow_Tables

The tools are designed to create a single table in many parts, because this is easier to do with the limits of GPU memory. The parts are merged together to create a single table. You repeat this process several times to generate several tables, which leads to your overall success rate.

Ok, something is clearing in my mind.. so if i create a chain of 200000 hashes, i can choose in how many parts it will be splitted? If i chose a higher number of tables parts, each of them will have only a smaller dimension and will contain a part of the chain of 200000. Is it right?

Each chain is atomic. It contains a start point and an end point. Each table contains many chains of the specified length. I don't know what you're trying to convey here, but I'm fairly sure it's fully wrong.

[Bitweasil]

My length 6 and 7 tables are being seeded as a bittorrent download. You may wish to start with those instead of trying to create your own.

http://cryptohaze.com/gpurainbowtables.php

[schnaps]

Hi, Bitweasil!

First of all, thanks for developing such a tool and for sharing it with the world.

I would like to see a 32-bit version, mostly because of the server part. I'm trying to configure a distributed setup, but the server I want to use to distribute the WU's is 32-bit (no graphic card, I intend to run --serveronly mode).

I also would like to leave a suggestion here: To be able to see current WU's progress and connected grafics + speeds on the client side. I assume this wasn't done to minimize bandwidth usage, maybe it could be added as an option.

Lastly, and sorry in advance for the off-topic, but is it normal that my CPU is not working with the cracking? I have an Intel i5 2410m.

Awaiting for your reply.
Thank you!

[Bitweasil]

I would like to see a 32-bit version, mostly because of the server part. I'm trying to configure a distributed setup, but the server I want to use to distribute the WU's is 32-bit (no graphic card, I intend to run --serveronly mode).

That would be a sane case. I'll see what I can do. Windows builds are not my strong suit, though.

I also would like to leave a suggestion here: To be able to see current WU's progress and connected grafics + speeds on the client side. I assume this wasn't done to minimize bandwidth usage, maybe it could be added as an option.

Minimize bandwidth usage, and there's typically no reason for the clients to need to know server details. I am considering a remote console, though.

Lastly, and sorry in advance for the off-topic, but is it normal that my CPU is not working with the cracking? I have an Intel i5 2410m.

Are you use Cryptohaze-Multiforcer (I assume so)? This does not support CPU.

The new framework, in alpha right now, does support CPU (mostly via OpenCL), but is not reliable yet.

[schnaps]

Are you use Cryptohaze-Multiforcer (I assume so)? This does not support CPU.
The new framework, in alpha right now, does support CPU (mostly via OpenCL), but is not reliable yet.

Hey! Thank you for the fast answer. Yes, i'm using Cryptohaze-Multiforcer. I thought I read somewhere it supported CPU, but I must have misassociated. Thank you for the update.

Cheers!