Posts: 5
Threads: 1
Joined: Apr 2013
04-14-2013, 05:46 PM
(This post was last modified: 04-15-2013, 12:23 AM by uzomoshi.)
Sapphire 5970, ubuntu 12.10 amd64, kernel 3.5.0-27, catalyst 13.1
Code: oclHashcat-plus64.bin --gpu-loops 1024 --gpu-accel 160 -m 120 -a 3 -o test-out.txt test.txt ?a?a?a?a?a?a
after a couple of minutes of run I still get
Code: Speed.GPU.#1...: 1044.1M/s
Speed.GPU.#2...: 1044.6M/s
Speed.GPU.#*...: 2088.7M/s
HWMon.GPU.#1...: 99% Util, 81c Temp, N/A Fan
HWMon.GPU.#2...: 99% Util, 84c Temp, N/A Fan
The only uncommon thing I can think of it's I put the card on a secondary PCI-E slot that is 8x. Other than that I just followed the book.
Isn't it slow? From benchmarks, etc.. I expected about 3200M/s for salted sha1.
Thank you
Posts: 601
Threads: 18
Joined: Apr 2010
Posts: 5
Threads: 1
Joined: Apr 2013
Could you explain a little bit why you say it's normal? I read everywhere that 6990 is only slightly better than 5970. Golubev estimation is about 15% better. It looks even less in real benchmarks like this one by epixoip http://hashcat.net/forum/thread-1804-pos...l#pid10242. Benchmarks on oclhashcat-plus homepage say PC3 with one 6990 does 3080M/s sha1. How a 5970 doing 2080M/s can be normal?
Thank you
Posts: 2,936
Threads: 12
Joined: May 2012
first, you're comparing lite to plus. second, you're using a salted algorithm, not a raw algorithm. third, we have no idea how many hashes/unique salts you're cracking because you did not paste the full output.
if you want to get numbers that you can actually compare, run:
Code: ./oclHashcat-lite64.bin -b --benchmark-mode 1
for plus, download the benchmark script at http://bindshell.nl/pub/hashcat-benchmark.sh
Posts: 2,936
Threads: 12
Joined: May 2012
also, i don't know if you caught it, but in my benchmark output that you link to, you'll see that i'm running my 5970s at 830Mhz. so if you're running yours at 725Mhz, you need to take that into account as well.
Posts: 5
Threads: 1
Joined: Apr 2013
Hi epixoip, your rig is outstanding! Yes I've considered everything you just mentioned. I know lite is faster and never meant to compare it to plus. Just compare lite results with lite results, like 6990 vs 5970 in your benchmarks. I've looked around and, also in your benchmarks, it looks like salt doesn't really make a difference in speed. I didn't try myself though. About overclocking, I tried 825/1010 and got less than 2500M/s. Still far from 3080M/s sha1 with 5970 stock clock mentioned in plus homepage. About hashes, my test was 200k generated sha1, single salt. But noticing it made a difference, I tried single hash single salt stock clock, getting about 2500M/s.
Later, I will try to benchmark the card properly and post it here.
Thank you
Posts: 2,936
Threads: 12
Joined: May 2012
the salt makes a difference depending on the number of hashes, and also whether it's salt.pass or pass.salt. i have only posted benchmarks for lite, which only supports pass.salt. -m 120 is salt.pass, which will be slower.
proper benchmarks should show you the speeds you expect.
Posts: 2,936
Threads: 12
Joined: May 2012
for reference, here's the speeds i get with 2x 5970
Code: Hash.Type......: MD5
Speed.GPU.#1...: 3665.4M/s
Speed.GPU.#2...: 3665.5M/s
Speed.GPU.#3...: 3665.4M/s
Speed.GPU.#4...: 3665.4M/s
Speed.GPU.#*...: 14661.7M/s
Hash.Type......: Joomla
Speed.GPU.#1...: 3276.7M/s
Speed.GPU.#2...: 3276.6M/s
Speed.GPU.#3...: 3276.7M/s
Speed.GPU.#4...: 3276.7M/s
Speed.GPU.#*...: 13106.7M/s
Hash.Type......: SHA1
Speed.GPU.#1...: 1670.7M/s
Speed.GPU.#2...: 1670.7M/s
Speed.GPU.#3...: 1670.7M/s
Speed.GPU.#4...: 1670.7M/s
Speed.GPU.#*...: 6682.8M/s
Hash.Type......: MSSQL(2000)
Speed.GPU.#1...: 1411.3M/s
Speed.GPU.#2...: 1411.6M/s
Speed.GPU.#3...: 1411.4M/s
Speed.GPU.#4...: 1411.4M/s
Speed.GPU.#*...: 5645.7M/s
Hash.Type......: NTLM..
Speed.GPU.#1...: 4751.8M/s
Speed.GPU.#2...: 4751.8M/s
Speed.GPU.#3...: 4751.8M/s
Speed.GPU.#4...: 4751.7M/s
Speed.GPU.#*...: 19007.1M/s
Hash.Type......: DCC, mscashCredentials, mscash...
Speed.GPU.#1...: 2744.3M/s
Speed.GPU.#2...: 2744.4M/s
Speed.GPU.#3...: 2744.3M/s
Speed.GPU.#4...: 2744.4M/s
Speed.GPU.#*...: 10977.5M/s
Hash.Type......: SHA256
Speed.GPU.#1...: 532.5M/s
Speed.GPU.#2...: 531.1M/s
Speed.GPU.#3...: 532.2M/s
Speed.GPU.#4...: 532.4M/s
Speed.GPU.#*...: 2128.2M/s
Hash.Type......: descrypt, DES(Unix), Traditional DES
Speed.GPU.#1...: 34186.3k/s
Speed.GPU.#2...: 34189.4k/s
Speed.GPU.#3...: 34189.4k/s
Speed.GPU.#4...: 34185.8k/s
Speed.GPU.#*...: 136.8M/s
Hash.Type......: SHA512
Speed.GPU.#1...: 69846.2k/s
Speed.GPU.#2...: 69641.9k/s
Speed.GPU.#3...: 69860.9k/s
Speed.GPU.#4...: 69826.9k/s
Speed.GPU.#*...: 279.2M/s
Hash.Type......: LM
Speed.GPU.#1...: 415.8M/s
Speed.GPU.#2...: 415.8M/s
Speed.GPU.#3...: 415.7M/s
Speed.GPU.#4...: 415.8M/s
Speed.GPU.#*...: 1663.3M/s
Hash.Type......: Oracle 7-10gES(Oracle)...
Speed.GPU.#1...: 69077.6k/s
Speed.GPU.#2...: 69018.8k/s
Speed.GPU.#3...: 69041.0k/s
Speed.GPU.#4...: 69051.0k/s
Speed.GPU.#*...: 276.2M/s
Benchmark complete.
Posts: 5
Threads: 1
Joined: Apr 2013
This is my benchmark 825/1010. I had to set --cpu-accel=240 instead of 800 because every run was killed. I guess due to insufficient memory (I got 1Gb ram). Results are comparable though and the small difference may be well due to cpu-accel.
So you were right, of course! salt+pass accounts to about 7% penalty, 200k hashes account to about 17% penalty and they add up to about 25% penalty I was observing.
Concluding, can you point out how to maximize a sha1 salt+pass attack?
Lite is not an option, I also tried the trick of putting salt in the mask (-m 100 ... <salt>?a?a?a?a?a), hoping -m 120 was optimized for many salts, but it just runs 10 times slower. 830 looks like the top without overvolting (and I don't want to). So, the only try looks like going -n 800. Do you know if I need more ram and suggest how much?
Thank you!!
Code: Hash.Type......: MD5
Speed.GPU.#1...: 3437.1M/s
Speed.GPU.#2...: 3475.8M/s
Speed.GPU.#*...: 6912.8M/s
Hash.Type......: Joomla
Speed.GPU.#1...: 3009.1M/s
Speed.GPU.#2...: 3064.6M/s
Speed.GPU.#*...: 6073.7M/s
Hash.Type......: SHA1
Speed.GPU.#1...: 1559.2M/s
Speed.GPU.#2...: 1569.4M/s
Speed.GPU.#*...: 3128.7M/s
Hash.Type......: MSSQL(2000)
Speed.GPU.#1...: 1161.6M/s
Speed.GPU.#2...: 1174.0M/s
Speed.GPU.#*...: 2335.6M/s
Hash.Type......: NTLM..
Speed.GPU.#1...: 4474.3M/s
Speed.GPU.#2...: 4499.2M/s
Speed.GPU.#*...: 8973.5M/s
Hash.Type......: DCC, mscashCredentials, mscash...
Speed.GPU.#1...: 2092.9M/s
Speed.GPU.#2...: 2131.6M/s
Speed.GPU.#*...: 4224.5M/s
Hash.Type......: SHA256
Speed.GPU.#1...: 462.8M/s
Speed.GPU.#2...: 467.7M/s
Speed.GPU.#*...: 930.5M/s
Hash.Type......: descrypt, DES(Unix), Traditional DES
Speed.GPU.#1...: 33932.2k/s
Speed.GPU.#2...: 33929.6k/s
Speed.GPU.#*...: 67861.8k/s
Hash.Type......: SHA512
Speed.GPU.#1...: 68280.5k/s
Speed.GPU.#2...: 68299.7k/s
Speed.GPU.#*...: 136.6M/s
Hash.Type......: LM
Speed.GPU.#1...: 411.8M/s
Speed.GPU.#2...: 412.8M/s
Speed.GPU.#*...: 824.6M/s
Hash.Type......: Oracle 7-10gES(Oracle)...
Speed.GPU.#1...: 68937.7k/s
Speed.GPU.#2...: 68014.2k/s
Speed.GPU.#*...: 137.0M/s
and may I add specific sha1 benchmarks
Code: Hash.Type......: SHA1
Speed.GPU.#1...: 1559.2M/s
Speed.GPU.#2...: 1569.4M/s
Speed.GPU.#*...: 3128.7M/s
Hash.Type......: SHA1 (salt + pass)
Speed.GPU.#1...: 1450.4M/s
Speed.GPU.#2...: 1450.3M/s
Speed.GPU.#*...: 2900.7M/s
Hash.Type......: SHA1 (200k hashes)
Speed.GPU.#1...: 1294.3M/s
Speed.GPU.#2...: 1296.4M/s
Speed.GPU.#*...: 2590.7M/s
Hash.Type......: SHA1 (salt + pass, 200k hashes)
Speed.GPU.#1...: 1175.1M/s
Speed.GPU.#2...: 1175.4M/s
Speed.GPU.#*...: 2350.5M/s
Posts: 5,185
Threads: 230
Joined: Apr 2010
I dont get what the problem is here. The numbers shown are the ones to expect.
|
