Posts: 42
Threads: 4
Joined: Sep 2014
02-25-2015, 05:55 PM
(This post was last modified: 02-25-2015, 05:57 PM by Quest.)
Hello, is that normal?..
===============================CudaHashcat 1.31:
Session.Name...: cudaHashcat
Status.........: Exhausted
Input.Base.....: Mask (?a?a?a?a) [4]
Input.Mod......: File (example.dict)
Hash.Target....: File (example0.hash)
Hash.Type......: MD5
Time.Started...: 0 secs
Time.Estimated.: 0 secs
Speed.GPU.#1...: 41186.6 kH/s
Recovered......: 2190/6494 (33.72%) Digests, 0/1 (0.00%) Salts
Progress.......: 136302297088/136302297088 (100.00%)
Skipped........: 0/136302297088 (0.00%)
Rejected.......: 0/136302297088 (0.00%)
HWMon.GPU.#1...: -1% Util, 65c Temp, -1% Fan
Started: Thu Feb 19 22:09:59 2015
Stopped: Thu Feb 19 22:24:07 2015 <---- = 14mins
===============================CudaHashcat 1.33:
Session.Name...: cudaHashcat
Status.........: Exhausted
Input.Left.....: Mask (?a?a?a?a) [4]
Input.Right....: File (example.dict)
Hash.Target....: File (example0.hash)
Hash.Type......: MD5
Time.Started...: Tue Feb 24 23:59:16 2015 (27 mins, 22 secs)
Time.Estimated.: 0 secs
Speed.GPU.#1...: 83355.9 kH/s <--- wut???
Recovered......: 2190/6494 (33.72%) Digests, 0/1 (0.00%) Salts
Progress.......: 136302297088/136302297088 (100.00%)
Skipped........: 0/136302297088 (0.00%)
Rejected.......: 0/136302297088 (0.00%)
HWMon.GPU.#1...: -1% Util, 71c Temp, -1% Fan
Started: Tue Feb 24 23:59:16 2015
Stopped: Wed Feb 25 00:26:49 2015 <---- = 27mins
??
NVIDIA GeForce GT 630M, CUDA 6.5.14, Nvidia 346.35,
Posts: 5,185
Threads: 230
Joined: Apr 2010
Sounds good for me (with v1.33). 136302297088 / 83355000 = 27 mins
Posts: 42
Threads: 4
Joined: Sep 2014
Hi atom,
the two benchmarks above were done with the exact same hardware/software setup, five days appart. I uninstalled 1.31 and installed 1.33.
How can it take twice as long with 1.33?
@ all, am I the only one with such results? If so nevermind 
Posts: 2,301
Threads: 11
Joined: Jul 2010
02-26-2015, 06:43 PM
(This post was last modified: 02-26-2015, 07:31 PM by undeath.)
I can confirm this problem:
cudaExample0.sh
cudaHashcat 1.31 @ 750Ti
Started: Thu Feb 26 17:39:51 2015
Stopped: Thu Feb 26 17:41:29 2015
>>> 98s
cudaHashcat 1.32b16 @ 750Ti
Started: Thu Feb 26 17:26:28 2015
Stopped: Thu Feb 26 17:29:41 2015
>>> 193s
cudaHashcat 1.34b7 @ 750Ti
Started: Thu Feb 26 17:30:42 2015
Stopped: Thu Feb 26 17:33:27 2015
>>> 165s
Posts: 42
Threads: 4
Joined: Sep 2014
02-26-2015, 06:58 PM
(This post was last modified: 02-26-2015, 07:00 PM by Quest.)
Thx undeath,
I just ran my test again with 1.31 then 1.33 with the same results.
This time I noticed something different in the beginning of the benchmark session... maybe it's nothing.
=========================1.31
root@The-Distribution-Which-Does-Not-Handle-OpenCL-Well (Kali):~# optirun cudahashcat64 -t 32 -a 7 example0.hash ?a?a?a?a example.dict
cudaHashcat v1.31 starting...
Enter YES in uppercase if you accept this EULA: YES
Device #1: GeForce GT 630M, 2047MB, 950Mhz, 2MCU
Device #1: WARNING! Kernel exec timeout is not disabled, it might cause you errors of code 702
Hashes: 6494 hashes; 6494 unique digests, 1 unique salts
Bitmaps: 16 bits, 65536 entries, 0x0000ffff mask, 262144 bytes
Applicable Optimizers:
* Zero-Byte
* Precompute-Init
* Precompute-Merkle-Demgard
* Meet-In-The-Middle
* Early-Skip
* Not-Salted
* Not-Iterated
* Single-Salt
* Scalar-Mode
* Raw-Hash
Watchdog: Temperature abort trigger set to 90c
Watchdog: Temperature retain trigger set to 80c
Device #1: Kernel ./kernels/4318/m00000_a1.sm_21.64.ptx
Device #1: Kernel ./kernels/4318/markov_le_v4.64.ptx
Device #1: Kernel ./kernels/4318/bzero.64.ptx <----------------------There
Generated dictionary stats for example.dict: 1210228 bytes, 129988 words, 129988 keyspace
=========================1.33
root@The-Distribution-Which-Does-Not-Handle-OpenCL-Well (Kali):~# optirun cudahashcat64 -t 32 -a 7 example0.hash ?a?a?a?a example.dict
cudaHashcat v1.33 starting...
Enter YES in uppercase if you accept this EULA: YES
Device #1: GeForce GT 630M, 2047MB, 950Mhz, 2MCU
Device #1: WARNING! Kernel exec timeout is not disabled, it might cause you errors of code 702
Hashes: 6494 hashes; 6494 unique digests, 1 unique salts
Bitmaps: 16 bits, 65536 entries, 0x0000ffff mask, 262144 bytes
Applicable Optimizers:
* Zero-Byte
* Precompute-Init
* Precompute-Merkle-Demgard
* Meet-In-The-Middle
* Early-Skip
* Not-Salted
* Not-Iterated
* Single-Salt
* Scalar-Mode
* Raw-Hash
Watchdog: Temperature abort trigger set to 90c
Watchdog: Temperature retain trigger set to 80c
Device #1: Kernel ./kernels/4318/m00000_a1.sm_21.64.ptx
Device #1: Kernel ./kernels/4318/markov_le_v4.64.ptx
<---------------------------------------------------->Not there
Generating dictionary stats for example.dict: 1210228 bytes (100.00%), 129988 words, 136302297088 ke
Generated dictionary stats for example.dict: 1210228 bytes, 129988 words, 136302297088 keyspace
Posts: 5,185
Threads: 230
Joined: Apr 2010
02-26-2015, 08:13 PM
(This post was last modified: 02-26-2015, 08:14 PM by atom.)
Here's the log that I had with undeath, it explains it in detail:
Quote:18:28:25 undeath | atom: it really seems something messed up from 1.31 to 1.32, cudaExample0 is almost twice as slow
18:30:56 @atom | nope its just optimized for fast systems
18:31:04 @atom | on fast ones its nearly twice as fast
18:32:03 undeath | what is a "fast" system?
18:32:27 @atom | well, a 7970 for example
18:32:33 @atom | nothing crazy
18:33:05 undeath | but if it's almost twice as slow with a 750ti something sounds off
18:40:48 @atom | root@et:~/xy/oclHashcat-1.31# ./oclHashcat64.bin -t 32 -a 7 example0.hash ?a?a?a?a example.dict --keyspace
18:40:48 @atom | oclHashcat v1.31 starting...
18:40:49 @atom | 1048576
18:40:55 @atom | root@et:~/oclHashcat-1.34# ./oclHashcat64.bin -t 32 -a 7 example0.hash ?a?a?a?a example.dict --keyspace
18:40:55 @atom | oclHashcat v1.34 starting...
18:40:56 @atom | 129988
18:41:02 @atom | thats the difference
18:41:26 @atom | in -a 7 mode, in 1.31, it used the mask as base
18:41:36 @atom | so it can parallel more
18:41:53 @atom | that changed with 1.32, because if you do -a 7 attack, you usually use bigger dictionaries
18:42:10 undeath | mhm, but still, 96s vs 165s is pretty hefty
18:42:22 @atom | no
18:42:23 undeath | 1.33 was even slower than that with 193s
18:42:44 @atom | its just hypothetical#
18:42:49 @atom | in reality, 1.33 will be faster
18:43:01 undeath | that was reality >.>
18:43:13 @atom | no its an example script
18:43:18 undeath | 
18:43:35 undeath | let me try a bigger dict
18:44:29 @atom | rockyou should be sufficient
18:45:03 undeath | yeah, just started with rockyou
18:45:15 @atom | you can even stick to example.dict
18:45:21 @atom | just remove 1 ?a from the mask
18:45:24 @atom | so only ?a?a?a
IOW, replace example.dict with rockyou and everything is fine
Posts: 42
Threads: 4
Joined: Sep 2014
02-26-2015, 08:21 PM
(This post was last modified: 02-26-2015, 08:39 PM by Quest.)
so if someone like me has a slower system, it's better to stick with 1.31 then?
Edit: Just noticed your comment at the bottom. Ok will try that! Thanks!
Posts: 5,185
Threads: 230
Joined: Apr 2010
Posts: 42
Threads: 4
Joined: Sep 2014
If anyone ran comparative benchmark(any) between 1.31 and 1.33 I would like to see those results. Just curious to see what you all get =]
|
