GET EXCITED! 1.2 rel vs 1.31 rc2
#1
Howdy folks.

Here's a benchmark of cudahc v1.2 'release' versus cudahc v1.31 'rc2' on GTX Titan @ 1.045 GHz.
Due to a discovery made by Jens, which resulted in an optimization for both NV and AMD GPUs, many algos got a decent boost in speed.

Top boosted algos:
SHA256(crypt) (+224%)
MD5(crypt) (+176%)
HMAC-SHA512 (+177%)
HMAC-SHA256 (+110%)
EPiS >4 (+109%)
DNSSEC (+106%)
DES(crypt) (+105%)

A bunch of algos lost a decent amount of speed (HMAC-SHA1 lost 84%, but Jens told me this may be a bug), full info below.

Code:
W:\cudaHashcat-1.31-rc2\cudaHashcat-1.31> W:\cudaHashcat-1.20\cudaHashcat-1.20> Device #1: GeForce GTX TITAN, 6144MB, 928Mhz, 14MCU Device #1: GeForce GTX TITAN, 6144MB, 928Mhz, 14MCU Hashtype: MD5 Hashtype: MD5 Speed.GPU.#1.: 7304.0 MH/s Speed.GPU.#1.: 4900.1 MH/s delta: +49.05% Hashtype: SHA1 Hashtype: SHA1 Speed.GPU.#1.: 1747.6 MH/s Speed.GPU.#1.: 2220.8 MH/s delta: -21.30% Hashtype: SHA256 Hashtype: SHA256 Speed.GPU.#1.: 619.7 MH/s Speed.GPU.#1.: 379.1 MH/s delta: +63.46% Hashtype: SHA512 Hashtype: SHA512 Speed.GPU.#1.: 147.4 MH/s Speed.GPU.#1.: 80356.7 kH/s delta: +83.43% Hashtype: SHA-3(Keccak) Hashtype: SHA-3(Keccak) Speed.GPU.#1.: 219.9 MH/s Speed.GPU.#1.: 118.2 MH/s delta: +86.04% Hashtype: SAP CODVN B (BCODE) Hashtype: SAP CODVN B (BCODE) Speed.GPU.#1.: 376.4 MH/s Speed.GPU.#1.: 308.9 MH/s delta: +21.85 Hashtype: SHA-1(Base64), nsldap, Netscape LDAP SHA Speed.GPU.#1.: 1747.9 MH/s Speed.GPU.#1.: 2220.5 MH/s delta: -21.28 Hashtype: SSHA-1(Base64), nsldaps, Netscape LDAP SSHA Speed.GPU.#1.: 1744.2 MH/s Speed.GPU.#1.: 1505.0 MH/s delta: +15.89 Hashtype: descrypt, DES(Unix), Traditional DES Speed.GPU.#1.: 72239.6 kH/s Speed.GPU.#1.: 35149.3 kH/s delta: +105.52% Hashtype: md5crypt, MD5(Unix), FreeBSD MD5, Cisco-IOS MD5 Speed.GPU.#1.: 3202.3 kH/s Speed.GPU.#1.: 1157.3 kH/s delta: +176.70% Hashtype: sha256crypt, SHA256(Unix) Hashtype: sha256crypt, SHA256(Unix) Speed.GPU.#1.: 71855 H/s Speed.GPU.#1.: 22174 H/s delta: +224.05% Hashtype: sha512crypt, SHA512(Unix) Hashtype: sha512crypt, SHA512(Unix) Speed.GPU.#1.: 26750 H/s Speed.GPU.#1.: 10831 H/s delta: +146.97% Hashtype: bcrypt, Blowfish(OpenBSD) Hashtype: bcrypt, Blowfish(OpenBSD) Speed.GPU.#1.: 896 H/s Speed.GPU.#1.: 957 H/s delta: -6.3% Hashtype: LM Hashtype: LM Speed.GPU.#1.: 839.9 MH/s Speed.GPU.#1.: 566.8 MH/s delta: +48.18% Hashtype: NTLM Hashtype: NTLM Speed.GPU.#1.: 12373.9 MH/s Speed.GPU.#1.: 11014.3 MH/s delta: +12.34% Hashtype: DCC, mscash Hashtype: DCC, mscash Speed.GPU.#1.: 3362.0 MH/s Speed.GPU.#1.: 2790.7 MH/s delta: +20.47% Hashtype: NetNTLMv1-VANILLA / NetNTLMv1+ESS Speed.GPU.#1.: 3334.8 MH/s Speed.GPU.#1.: 2520.6 MH/s delta: +32.30% Hashtype: NetNTLMv2 Hashtype: NetNTLMv2 Speed.GPU.#1.: 284.9 MH/s Speed.GPU.#1.: 322.9 MH/s delta: -11.76% Hashtype: EPiServer 6.x < v4 Hashtype: EPiServer 6.x < v4 Speed.GPU.#1.: 643.8 MH/s Speed.GPU.#1.: 1758.6 MH/s delta: -63.39% Hashtype: EPiServer 6.x > v4 Hashtype: EPiServer 6.x > v4 Speed.GPU.#1.: 591.3 MH/s Speed.GPU.#1.: 282.3 MH/s delta: +109.45% Hashtype: MSSQL(2000) Hashtype: MSSQL(2000) Speed.GPU.#1.: 1755.8 MH/s Speed.GPU.#1.: 1496.2 MH/s delta: +17.35% Hashtype: MSSQL(2005) Hashtype: MSSQL(2005) Speed.GPU.#1.: 1755.6 MH/s Speed.GPU.#1.: 1497.5 MH/s delta: +17.23% Hashtype: MSSQL(2012) Hashtype: MSSQL(2012) Speed.GPU.#1.: 149.6 MH/s Speed.GPU.#1.: 80277.4 kH/s delta: +86.35% Hashtype: MySQL323 Hashtype: MySQL323 Speed.GPU.#1.: 18473.9 MH/s Speed.GPU.#1.: 9653.0 MH/s delta: +91.37% Hashtype: Oracle 7-10g Hashtype: Oracle 7-10g Speed.GPU.#1.: 174.0 MH/s Speed.GPU.#1.: 110.6 MH/s delta: +57.32% Hashtype: Sybase ASE Hashtype: Sybase ASE Speed.GPU.#1.: 47525.3 kH/s Speed.GPU.#1.: 36255.9 kH/s delta: +31.08% Hashtype: OSX v10.4, v10.5, v10.6 Hashtype: OSX v10.4, v10.5, v10.6 Speed.GPU.#1.: 643.4 MH/s Speed.GPU.#1.: 1750.7 MH/s delta: -63.24% Hashtype: OSX v10.7 Hashtype: OSX v10.7 Speed.GPU.#1.: 148.0 MH/s Speed.GPU.#1.: 75087.2 kH/s delta: +97.10% Hashtype: OSX v10.8 / v10.9 Hashtype: OSX v10.8 / v10.9 Speed.GPU.#1.: 1874 H/s Speed.GPU.#1.: 1060 H/s delta: +76.79% Hashtype: Cisco-PIX MD5 Hashtype: Cisco-PIX MD5 Speed.GPU.#1.: 5049.2 MH/s Speed.GPU.#1.: 4231.9 MH/s delta: +19.31% Hashtype: Cisco-ASA MD5 Hashtype: Cisco-ASA MD5 Speed.GPU.#1.: 5042.5 MH/s Speed.GPU.#1.: 4217.7 MH/s delta: +19.55% Hashtype: Cisco-IOS SHA256 Hashtype: Cisco-IOS SHA256 Speed.GPU.#1.: 619.7 MH/s Speed.GPU.#1.: 380.3 MH/s delta: +62.95% Hashtype: DNSSEC (NSEC3) Hashtype: DNSSEC (NSEC3) Speed.GPU.#1.: 810.3 MH/s Speed.GPU.#1.: 393.0 MH/s delta: +106.1% Hashtype: WPA/WPA2 Hashtype: WPA/WPA2 Speed.GPU.#1.: 99550 H/s Speed.GPU.#1.: 112.0 kH/s delta: -11.11% Hashtype: IKE-PSK MD5 Hashtype: IKE-PSK MD5 Speed.GPU.#1.: 159.0 MH/s Speed.GPU.#1.: 349.3 MH/s delta: -54.48% Hashtype: 1Password, cloudkeychain Hashtype: 1Password, cloudkeychain Speed.GPU.#1.: 1637 H/s Speed.GPU.#1.: 928 H/s delta: +76.4% Hashtype: TrueCrypt 5.0+ PBKDF2-HMAC-SHA512 + AES Speed.GPU.#1.: 64397 H/s Speed.GPU.#1.: 36654 H/s delta: +75.68% Hashtype: Drupal7 Hashtype: Drupal7 Speed.GPU.#1.: 8983 H/s Speed.GPU.#1.: 4569 H/s delta: +96.6% Hashtype: HMAC-MD5 (key = $pass) Hashtype: HMAC-MD5 (key = $pass) Speed.GPU.#1.: 344.2 MH/s Speed.GPU.#1.: 610.3 MH/s delta: -43.6% Hashtype: HMAC-MD5 (key = $salt) Hashtype: HMAC-MD5 (key = $salt) Speed.GPU.#1.: 1704.7 MH/s Speed.GPU.#1.: 1154.2 MH/s delta: +47.69% Hashtype: HMAC-SHA1 (key = $pass) Hashtype: HMAC-SHA1 (key = $pass) Speed.GPU.#1.: 75191.3 kH/s Speed.GPU.#1.: 177.8 MH/s delta: -57.71% Hashtype: HMAC-SHA1 (key = $salt) Hashtype: HMAC-SHA1 (key = $salt) Speed.GPU.#1.: 139.0 MH/s Speed.GPU.#1.: 872.9 MH/s delta: -84.07% Hashtype: HMAC-SHA256 (key = $pass) Hashtype: HMAC-SHA256 (key = $pass) Speed.GPU.#1.: 129.2 MH/s Speed.GPU.#1.: 61299.1 kH/s delta: +110.76% Hashtype: HMAC-SHA256 (key = $salt) Hashtype: HMAC-SHA256 (key = $salt) Speed.GPU.#1.: 267.6 MH/s Speed.GPU.#1.: 144.3 MH/s delta: +85.44% Hashtype: HMAC-SHA512 (key = $pass) Hashtype: HMAC-SHA512 (key = $pass) Speed.GPU.#1.: 29651.9 kH/s Speed.GPU.#1.: 10671.4 kH/s delta: +177.86% Hashtype: HMAC-SHA512 (key = $salt) Hashtype: HMAC-SHA512 (key = $salt) Speed.GPU.#1.: 58030.8 kH/s Speed.GPU.#1.: 37228.9 kH/s delta: +55.87% Hashtype: IPMI2 RAKP HMAC-SHA1 Hashtype: IPMI2 RAKP HMAC-SHA1 Speed.GPU.#1.: 424.6 MH/s Speed.GPU.#1.: 814.9 MH/s delta: -47.89% Hashtype: Double MD5 Hashtype: Double MD5 Speed.GPU.#1.: 885.4 MH/s Speed.GPU.#1.: 1367.0 MH/s delta: -35.23% Hashtype: GRUB 2 Hashtype: GRUB 2 Speed.GPU.#1.: 6553 H/s Speed.GPU.#1.: 3704 H/s delta: +76.91% Hashtype: phpass, MD5(Wordpress), MD5(phpBB3), MD5(Joomla) Speed.GPU.#1.: 2033.0 kH/s Speed.GPU.#1.: 1430.0 kH/s delta: +42.16% Hashtype: Joomla < 2.5.18 Hashtype: Joomla < 2.5.18 Speed.GPU.#1.: 7285.8 MH/s Speed.GPU.#1.: 5144.8 MH/s delta: +41.61% Hashtype: osCommerce, xt:Commerce Hashtype: osCommerce, xt:Commerce Speed.GPU.#1.: 3983.9 MH/s Speed.GPU.#1.: 2768.8 MH/s delta: +43.88 Hashtype: IPB2+, MyBB1.2+ Hashtype: IPB2+, MyBB1.2+ Speed.GPU.#1.: 468.7 MH/s Speed.GPU.#1.: 929.3 MH/s delta: -49.56% Hashtype: vBulletin < v3.8.5 Hashtype: vBulletin < v3.8.5 Speed.GPU.#1.: 886.3 MH/s Speed.GPU.#1.: 1367.3 MH/s delta: -35.17% Hashtype: vBulletin > v3.8.5 Hashtype: vBulletin > v3.8.5 Speed.GPU.#1.: 448.7 MH/s Speed.GPU.#1.: 900.9 MH/s delta: -50.19% Hashtype: SMF > v1.1 Hashtype: SMF > v1.1 Speed.GPU.#1.: 645.5 MH/s Speed.GPU.#1.: 1752.0 MH/s delta: -63.15%
#2
Thanks Rolf for the comparison, good work!

For all others, keep in mind that those changes only reflect the change for GTX TITAN. Previous TITAN support wasn't perfect (as you can see).

There's of course performance increase for other NV cards as well, but they look very different.
#3
Can Maxwell, potentially, be tested using frankenstein kernels on 1.2?
I'm very interested in SM_52's results, because GTX 980 is now faster than vanilla Titan.
#4
That will not work (or at least very unfair) because of the newly added instructions that make Maxwell that fast. I'm sure CUDA runtime will complain