GRTGen is the tool used to generate table parts. It is designed to generate a large number of table parts that are merged - it is not designed to generate the entire table in one pass. It will put lots of parts in the "parts" directory which are then merged using the GRTMerge utility.
Usage
GRTGen-[CUDA | OpenCL] -h [NTLM | MD5 | SHA1] -c charsetfile -l <password length> -i
--chainlength <chain length> [options]
Options
Autotune: --autotune
This option is not currently used, and does nothing.
-b (--blocks) Number of thread blocks to run
--bits Bits in the table output
--bfi_int Use AMD gfi_int (OpenCL Only)
-c (--charsetfile) Path to the charset file
--chainlength Integer value for length of each chain
-d (--device) Device to use, default 0
-h (--hashtype) Set the hashtype for the generated rainbow table. Available hashes are NTLM, MD4, MD5, or SHA1
-i (--tableindex) Integer for the table's index
-l (--passwordlen) Integer password length must be greater than 4
-m (--ms) Target step time in ms
--numchains Integer value for number of chains in each table
--numtables Number of tables to create
-p (--platform) Platform to use, default 0
--seed Seed value to use, used for testing
-t (--threads) Number of threads per block
--tableversion Table version 1 or 2, default 2
--vectorwidth Vector width, default 4
Example
GRTGen-CUDA -h MD5 -c ./charsets/charsetall -l 6 -i 0 --chainlength 50000 --numchains 10000000 -t 128 -b 128 -m 500