[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 26 Sep 2012 21:15:39 +0400
From: Gifts <gifts.antichat@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: sha512crypt-opencl
Benchmark with 20 minutes JtR and 10 minutes cudaHashcat
gifts@...ts-arch:~/work/JohnTheRipper/run$ ./john
-format:sha512crypt-opencl -platform=1 --test
OpenCL platform 1: NVIDIA CUDA, 1 device(s).
Using device 0: GeForce GTX 560 Ti
Building the kernel, this could take a while
Local work size (LWS) 512, global work size (GWS) 4096
Benchmarking: sha512crypt (rounds=5000) [OpenCL]... DONE
Raw: 6884 c/s real, 6942 c/s virtual
gifts@...ts-arch:~/work/JohnTheRipper/run$ ./john
-format:sha512crypt-opencl -platform=1 -i=all8 /tmp/sha512.txt
OpenCL platform 1: NVIDIA CUDA, 1 device(s).
Using device 0: GeForce GTX 560 Ti
Building the kernel, this could take a while
Local work size (LWS) 512, global work size (GWS) 4096
Loaded 1 password hash (sha512crypt [OpenCL])
guesses: 0 time: 0:00:20:04 0.00% c/s: 7120 trying: cluegelo - clleam18
Session aborted
gifts@...ts-arch:~/work/hashcat/oclHashcat-plus-0.09$
./cudaHashcat-plus64.bin -a 3 -m 1800 -n 40 --gpu-loops 128
/tmp/sha512.txt ?l?l?l?l?l?l?l?l
cudaHashcat-plus v0.09 by atom starting...
Hashes: 1 total, 1 unique salts, 1 unique digests
Bitmaps: 8 bits, 256 entries, 0x000000ff mask, 1024 bytes
Workload: 128 loops, 40 accel
Watchdog: Temperature abort trigger set to 90c
Watchdog: Temperature retain trigger set to 80c
Device #1: GeForce GTX 560 Ti, 1023MB, 1645Mhz, 8MCU
Device #1: Kernel ./kernels/4318/m1800.sm_21.ptx
[s]tatus [p]ause [r]esume [b]ypass [q]uit => q
Status.......: Aborted
Input.Mode...: Mask (?l?l?l?l?l?l?l?l)
Hash.Target..: $6$UHyS5Yyx$.zgXTZm0TaLTYlgtxhOlILzWZ52x9KD5pHInyE7Jz4cNZlIvA.CzpKRwFoHoFUWS37KlNnOSaf1Xlpr49k5en/
Hash.Type....: sha512crypt, SHA512(Unix)
Time.Running.: 10 mins, 56 secs
Time.Left....: 1 year, 37 days
Time.Util....: 656746.2ms/148.3ms Real/CPU, 0.0% idle
Speed........: 6486 c/s Real, 6488 c/s GPU
Recovered....: 0/1 Digests, 0/1 Salts
Progress.....: 4259840/208827064576 (0.00%)
Rejected.....: 0/4259840 (0.00%)
HWMon.GPU.#1.: -1% Util, 73c Temp, 80% Fan
Started: Wed Sep 26 20:49:48 2012
Stopped: Wed Sep 26 21:00:44 2012
cudaHashCat might be accelerated a little with -n and --gpu-loops,
because with default values it shows ~5800 c/s. But further increasing
of this values makes system unresponsive, so I can't say what exact
values it would be.
And it is interesting, that Intel OpenCl SDK can't vectorize this
kernel. I hadn't look deeply into code yet, was this kernel optimized
for GPU?
2012/9/26 Claudio André <claudioandre.br@...il.com>:
> Em 26-09-2012 02:52, Solar Designer escreveu:
>
>> Claudio, Gifts -
>>
>> A while ago, atom let us know that his sha512crypt implementation in
>> oclHashcat-plus is "87% faster" than ours on NVIDIA's 560Ti. Gifts has
>> now benchmarked the two on his 560Ti, and at first glance this does not
>> appear to be the case - however, more testing is needed.
>
> Good to hear.
>
> How about the plans to include/apply the "self-test time impact" solution
> (by solar) to unstable? The self-test time hurts c/s A LOT.
>
> BTW: Thanks Gifts
>
>
>
>> https://gist.github.com/596c0b03c7ed78ef4937
>>
>> Gifts - can you please use -i=all8 for the actual run, and let it run
>> for 10 minutes? When you give no options, JtR uses "batch mode", which
>> includes a "single crack" mode phase. "Single crack" mode is
>> GPU-unfriendly. Besides, we currently include self-test duration in
>> total running time used for c/s rate calculation.
>>
>> Thanks,
>>
>> Alexander
>
>
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
