hashcat Forum

Full Version: oclHashcat v1.35
You're currently viewing a stripped down version of our content. View the full version with proper formatting.


Download here: https://hashcat.net/oclhashcat/



This version 1.35 is about performance increase and bug fixes. Even if you did not face any errors with v1.33 we recommend upgrading.

Good thing is both oclHashcat v1.35 and oclHashcat v1.33 share the same driver dependancies. If you had a running v1.33 then v1.35 will work, too.

Most important changes:
  • Increased speed for SHA1 raw kernels, all AMD GPU
  • Increased speed for PDF kernel, all NVidia GPU
  • Increased speed for all multihash kernel using a different bitmap technique, all GPU
  • Fixed a race condition in multihash kernels
  • Fixed a buffer overflow that could lead to all kinds of segmentation fault or system freezing
  • Fixed a bug which caused clCreateBuffer() -61 Error even on single hashes



Full changelog v1.34 -> v1.35

Quote:
Type: Feature
File: Kernel
Desc: Increased speed for all multihash kernel using a bitmap technique

Type: Bug
File: Kernels
Desc: Fixed a bug in NVidia multihash kernels that a hash was not found even with the correct plain

Full changelog v1.33 -> v1.34

Quote:
Type: Feature
File: Kernel
Desc: Increased speed for SHA1 raw kernels (-m 100/110/120/130/140) on GCN cards (7970, 280x, 290x, ...)

Type: Feature
File: Kernel
Desc: Increased speed for PDF kernel (-m 104xx and -m 105xx) for all NVidia GPU

Type: Feature
File: Kernel
Desc: Increased speed for all multihash kernel using a different bitmap technique

Type: Feature
File: Kernel
Desc: Added oclHashcat compile timestamp to buffered binary kernels to avoid collisions with older versions

Type: Feature
File: Kernel
Desc: Added new hash mode -m 11000 = PrestaShop
Trac: #589

Type: Feature
File: Kernel
Desc: Added new hash mode -m 11100 = PostgreSQL Challenge-Response Authentication (MD5)
Trac: #490

Type: Feature
File: Kernel
Desc: Added new hash mode -m 11200 = MySQL Challenge-Response Authentication (SHA1)
Trac: #490

Type: Feature
File: Kernel
Desc: Added support for 32 bytes ID objects for hash type -m 10500 = PDF 1.4 - 1.6
Trac: #594

Type: Feature
File: Host
Desc: Inform user with small wordlists how to avoid slow speed with slow hashes
Trac: #597

Type: Change
File: Host
Desc: Renamed -m 7100 = OS X v10.8 / v10.9 to OS X v10.8+

Type: Workaround
File: Host
Desc: New workaround for -m 2500 = WPA/WPA2 key versions which are not valid (should always be < 256)

Type: Bug
File: Kernels
Desc: Fixed a race condition in multihash kernels

Type: Bug
File: Kernels
Desc: Fixed a buffer overflow that could lead to all kinds of segmentation fault or system freezing
Trac: #591

Type: Bug
File: Host
Desc: The key size (AES 128 vs AES 256) for office 2007 hashes was not shown correctly in output/status
Trac: #590

Type: Bug
File: Host
Desc: Fixed a bug if a rule in -j changes the length of the words in the wordlist, ex. ]
Trac: #593

Type: Bug
File: Host
Desc: Fixed a crash when using the outfile remove feature together with some hash types
Trac: #593

Type: Bug
File: Host
Desc: Fixed a problem where wrong restore point percentage was shown
Trac: #606

Type: Bug
File: Host
Desc: Fixed a buffer overflow in stdin mode

Type: Bug
File: Host
Desc: Fixed a bug which caused clCreateBuffer() -61 Error even on single hashes

Type: Bug
File: Host
Desc: Fixed a bug which caused clEnqueueCopyBuffer() -30 Error by limiting --gpu-loops to a maximum of 1024

Type: Bug
File: Host
Desc: Fixed a bug related to the maximum password length supported per hash-type

--
atom
Rolf was here.

Code:
Type: Feature
File: Kernel
Desc: Added oclHashcat compile timestamp to buffered binary kernels to avoid collisions with older versions
No more frankencat?
Hm... if i use cudaHashcat64.exe -a 3 -m 20 -2 ?l?d -3 ?l?u hash.txt ?3?2?2?2?2?2?2
Quote:ATTENTION!
The wordlist or mask you are using is too small.
That mask is small?!
It need 4 hours to compute for 1000 hashes! (more hashes-more time and always one gpu)

And again i must use voodoo magic like --gpu-accel=2..6 or slow pipe Sad

Maybe be better if cudaHashcat64 can be "internally pipe" by itself for small mask? (without "external" pipe programm)
keyspace is only 52 + 36^6, that is incredibly small
Ok 4 hours by gtx970 is small... but this make compute too long, if i use mask file with small and big keyspace inside.
User must calculated each mask and use many different ways to compute?! I dont think that is right way Sad
Maybe be right if hashcat can make all it internally?

I hope that is be fixed https://hashcat.net/trac/ticket/605 Sad

How i can understand, mask small or big?
So I ran the exact same command on the same capture and same word list just as a comparison.

Code:
./oclHashcat64.bin -m 2500 -w 3 --gpu-temp-retain=60 --status ~/captures/netgear38.hccap ~/wordlists/SANITIZEDwpa2

Now I realize this is only against one hash type but the results were almost identical. Just providing info in case you wanted it. Let me know if you would like any other tests ran. But have updated to the new version and will update on anything useful. Thanks Atom!!!!! Great as always!

Code:
Session.Name...: oclHashcat
Status.........: Exhausted
Input.Mode.....: File (/home/dylan/wordlists/SANITIZEDwpa2)
Hash.Target....: NETGEAR38 (20:e5:2a:18:d1:ca <-> 80:86:f2:db:37:ba)
Hash.Type......: WPA/WPA2
Time.Started...: Sun Mar 22 11:30:22 2015 (1 hour, 46 mins)
Time.Estimated.: 0 secs
Speed.GPU.#1...:    66189 H/s
Speed.GPU.#2...:    91738 H/s
Speed.GPU.#*...:   157.9 kH/s
Recovered......: 0/1 (0.00%) Digests, 0/1 (0.00%) Salts
Progress.......: 982963903/982963903 (100.00%)
Skipped........: 0/982963903 (0.00%)
Rejected.......: 0/982963903 (0.00%)
HWMon.GPU.#1...: 91% Util, 52c Temp, 42% Fan
HWMon.GPU.#2...: 92% Util, 52c Temp, 80% Fan

Started: Sun Mar 22 11:30:22 2015
Stopped: Sun Mar 22 13:22:30 2015