GTX 480 Core

[blazer]

What is the speed when you go ?

./CUDA-Multiforcer -d 1 -l -h NTLM -f <FILE> --min=8 --max=8 -c charset/charsetfull -o <Output File> --threads 512 --blocks 15 -m 500

[Angel]

GTX480 box - ver 0.7 build from source!

NTLM (optimal perfomance - increasing --threads/--blocks/-m give me same result):

Code: Select all

# ./multiforcer -l -h NTLM -f Hashes-NTLM-Full.txt --min=8 --max=8 -c charsets/all --threads 512 --blocks 60 -m 500
...
Done: 0.01%  Step rate: 1711.5M/s Search rate: 23960.5M/sec


4 blazer with --blocks 15:

Code: Select all

# ./multiforcer -l -h NTLM -f Hashes-NTLM-Full.txt --min=8 --max=8 -c charsets/all --threads 512 --blocks 15 -m 500
...
Done: 0.01%  Step rate: 855.2M/s Search rate: 11972.6M/sec


MD5:

Code: Select all

# ./multiforcer --threads 512 --blocks 60 -m 500 -h MD5 -f test -c charsets/all --min=8 --max=8
Cryptohaze.com CUDA Multiforcer (multiple hash brute forcer)
by Bitweasil
Version 0.70, length 0-14
Currently supported hash types: MD5 FASTMD5 MD4 FASTMD4 NTLM FASTNTLM SHA1 FASTSHA1
Hash type: MD5
CUDA Device Information:
Device 0: "GeForce GTX 480"
  Number of cores:                               120
  Clock rate:                                    1.40 GHz
  Memory (total):                                1535 MB
  Memory (free):                                 1358 MB

Single charset loaded.
Loading & sorting hashes.  This may take a while.
Returning from sort.
Hashes loaded (1 hashes)
You may want to consider the FAST[hash] option if it exists.
Done with copy to constant
Launching kernel for password length 8
Done: 0.01%  Step rate: 1117.7M/s Search rate: 1117.7M/sec


very nice software
oclHashcat give almost same speed (but don't have sources so it suck)
bvernoux gpu_md5_crack give me 400M/s and seems can't be tuned
tnks

[re8el]

Hello,

my setup:
ubuntu 10.10 x64
nvidia 256.53
gpu toolkit
opencl 1.0

wen running
/CUDA-Multiforcer-Linux-0.72# ./CUDA-Multiforcer -d 1 -l -h NTLM -f test_hashes/Hashes-NTLM-Full.txt --min=8 --max=8 -c charsets/charsetlowernumeric -o output.file --threads 512 --blocks 60 -m 500

i get this error:

CUDA Device Information:
Device 1: ""
Number of cores: 0
Clock rate: 0.00 GHz
Performance Number: 0
Note: Performance number is clock in mhz * core count, for comparing devices.
Single charset loaded.
Loading & sorting hashes. This may take a while.
Hashes loaded (14 hashes)
You may want to consider the FAST[hash] option if it exists.
Cuda error: invalid device ordinal.

any ideia?

Best Regards
re8el

[Bitweasil]

You don't have CUDA installed, or it's not running correctly. Try running the multiforcer as root once, or start X11 - you may not have /dev/nvidia set up properly.

[Sc00bz]

It looks like you are trying to use a device that doesn't exist. -d specifies which device to use.
./CUDA-Multiforcer -d 1 -l -h NTLM -f test_hashes/Hashes-NTLM-Full.txt --min=8 --max=8 -c charsets/charsetlowernumeric -o output.file --threads 512 --blocks 60 -m 500

Try:
./CUDA-Multiforcer -l -h NTLM -f test_hashes/Hashes-NTLM-Full.txt --min=8 --max=8 -c charsets/charsetlowernumeric -o output.file --threads 512 --blocks 60 -m 500

[re8el]

Hi

Sc00bz, this time it run.

Weird with no -d or with -d 0 it runs ok. I have the idea that i could use all the 480 cores of the board, but it's using only 120 !?!??!

Is this results what to be expected from this card?

Code: Select all

cuda-tools/CUDA-Multiforcer-Linux-0.72# ./CUDA-Multiforcer -l -h NTLM -f test_hashes/Hashes-NTLM-Full.txt --min=8 --max=8 -c charsets/charsetlowernumeric -o output.file --threads 512 --blocks 60 -m 500
Cryptohaze.com CUDA Multiforcer (multiple hash brute forcer)
by Bitweasil
Version 0.72, length 0-14
Currently supported hash types: MD5 FASTMD5 MD4 FASTMD4 NTLM FASTNTLM SHA1 FASTSHA1
Hash type: NTLM
CUDA Device Information:
Device 0: "GeForce GTX 480"
  Number of cores:                               120
  Clock rate:                                    1.51 GHz
  Performance Number:                            22680
  Note: Performance number is clock in mhz * core count, for comparing devices.
Single charset loaded.
Loading & sorting hashes.  This may take a while.
Hashes loaded (14 hashes)
You may want to consider the FAST[hash] option if it exists.
Launching kernel for password length 8
Done: 19.20%  Step rate: 1669.7M/s Search rate: 23375.5M/sec
NTLM:8846F7EAEE8FB117AD06BDD830B7586C:password:0x70617373776f7264
Done: 99.99%  Step rate: 1663.7M/s Search rate: 23291.6M/sec

------------------------------------------

Compute done: Reference time 1697.1 seconds
Stepping rate: 1662.3M MD4/s
Search rate: 23272.6M NTLM/s

[Bitweasil]

I have no idea why it's only seeing 120 cores. You have the latest drivers installed?

I currently don't have any Fermi devices, so I can't really test it on them. The *theory* of the donate button on the front page was to cover some of my hardware costs & fund GPUs, but that's not working so well.

[Sc00bz]

It's definitely using all 480 cores because it's getting a step rate of 1663.7M/s. So don't worry the only thing wrong is the status message. It looks like the performance number is wrong too.

Code: Select all

  Number of cores:                               120
  Clock rate:                                    1.51 GHz
  Performance Number:                            22680
  Note: Performance number is clock in mhz * core count, for comparing devices.
Single charset loaded.

1,510 * 120 = 181,200 != 22,680

9800 GTX+:

Code: Select all

  Number of cores:                               128
  Clock rate:                                    1.84 GHz
  Performance Number:                            29376

1,840 * 128 = 235,520 != 29,376

[re8el]

I have the driver 256.53 with cudatoolkit 3.1 ubuntu9.10 with grucompsdk 3.1

It's running on a ubuntu 10.10 x64 and its a zotac gtx 480 amp.

this is really wierd...

[Bitweasil]

I'll take a look at that. I might be wrapping it into a char or something.

... anyone got a shell box with a 480 in it I can use?