Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Thu, 22 Nov 2012 18:49:34 +0100
From: groszek <necro@...icon.pl>
To: john-users@...ts.openwall.com
Subject: Re: What do you recommend as a laptop cracking station?
 :)

On 11/22/2012 07:05 AM, wfdawson wrote:
> I've been thinking about a new system for myself, also.  What spells Xmas like crushing credit card debt, after all. :-)  
> 
> Not having my own system newer than a 10 year old workstation, I'm using a work-supplied laptop (Dell E6410, i5, 8GB RAM, SAMSUNG SSD 830) for cracking.  It's hardly the sort of thing to brag about.  Indeed, heat is my biggest enemy.  I've long since blown out the heat sink / fan on this loser.  It only manages to keep limping along with the USB powered laptop cooler attached.  I suspect the same would be true for  other laptops as well.  My next company supplied laptop will be a Lenovo, though not a y580 (sadly).
> 
> It may be worthwhile to compare numbers to get an idea of what the latest, greatest laptop hardware vs. an older model buys one.  Incidentally, I'm using Pentoo from portage overlay (not the live builds), but updated more or less daily.  The latest available (test) build from Gentoo for john is "1.7.9-r5", though john reports somewhat differently.
> 
> Strangely enough, this system outperforms the Lenovo y580 on the CPU (not CUDA, obviously) benchmarks for M$ Cache Hash MD4 and NTLMv2 C/R MD4 HMAC-MD5 runs.  Odd, what?  I didn't see additional hardware details like RAM and SSD vs. HDD mentioned, but those will obviously have some impact on overall performance.  On the other hand, they're "just" benchmarks so have to be viewed with some distrust, anyway.
> 
> ver: 1.7.9-jumbo-7_omp [linux-x86-64]
> 
> Benchmarking: M$ Cache Hash MD4 [32/64]... (4xOMP) DONE
> Many salts:     37584K c/s real, 9396K c/s virtual
> Only one salt:  13199K c/s real, 3308K c/s virtual
> 
> Benchmarking: M$ Cache Hash 2 (DCC2) PBKDF2-HMAC-SHA-1 [128/128 SSE2 intrinsics 8x]... (4xOMP) DONE
> Raw:    355 c/s real, 89.0 c/s virtual
> 
> Benchmarking: MS SQL SHA-1 [128/128 SSE2 intrinsics 8x]... DONE
> Many salts:     4124K c/s real, 4044K c/s virtual
> Only one salt:  3460K c/s real, 3495K c/s virtual
> 
> Benchmarking: NTLMv2 C/R MD4 HMAC-MD5 [32/64]... (4xOMP) DONE
> Many salts:     2228K c/s real, 555666 c/s virtual
> Only one salt:  1720K c/s real, 431157 c/s virtual
> 
> 
> Using latest unstable build from git (ver: 1.7.9-jumbo-7+unstable_omp [linux-x86-64i]), the results are comparable, except now my i5 outperforms the y580 CPU on the M$ Cache Hash 2 (DCC2) PBKDF2-HMAC-SHA-1 run also.
> 
> 
> Benchmarking: M$ Cache Hash MD4 [32/64]... (4xOMP) DONE
> Many salts:     37953K c/s real, 9512K c/s virtual
> Only one salt:  13066K c/s real, 3266K c/s virtual
> 
> 
> Benchmarking: M$ Cache Hash 2 (DCC2) PBKDF2-HMAC-SHA-1 [128/128 SSE2 intrinsics 8x]... (4xOMP) DONE
> Raw:    1312 c/s real, 327 c/s virtual
> 
> Benchmarking: MS SQL SHA-1 [128/128 SSE2 intrinsics 8x]... DONE
> Many salts:     14171K c/s real, 14171K c/s virtual
> Only one salt:  8838K c/s real, 8928K c/s virtual
> 
> Benchmarking: NTLMv2 C/R MD4 HMAC-MD5 [32/64]... (4xOMP) DONE
> Many salts:     2189K c/s real, 552960 c/s virtual
> Only one salt:  1720K c/s real, 431157 c/s virtual
> 
> 
> My personal interest of the moment (and the past 12 months or so) is with Lotus Domino unsalted and salted hashes.
> 
> Benchmarking: Lotus Notes/Domino 6 More Secure Internet Password [8/64]... DONE
> Many salts:     117593 c/s real, 117593 c/s virtual
> Only one salt:  69995 c/s real, 69995 c/s virtual
> 
> Benchmarking: Lotus Notes/Domino 5 [8/64]... (4xOMP) DONE
> Raw:    594432 c/s real, 148608 c/s virtual
> 
> 
> I'd like to see benchmark results for these two hash format and hardware info from other platforms, if anyone is willing to oblige.
> 
> Thanks!
> 
> 

I see you have used 4xOMP - I should've noted I ran the cpu benchmarks
on just one core. Silly of me I know, since we're talking about what
speed we can actually pull out of those machines .. That would explain
the differences. Since the tests are being run entirely in memory the
disk speed doesn't change anything. But for the record its a standard
laptop 5400 rpm drive, and I have 8 gb ddr3 ram.

Here's the lotus test (on 1/4/8 cores/threads):

Benchmarking: Lotus Notes/Domino 5 [8/64]... DONE
Raw:    451675 c/s real, 456192 c/s virtual

Benchmarking: Lotus Notes/Domino 5 [8/64]... (4xMPI) DONE
Raw:    1704K c/s real, 1721K c/s virtual

Benchmarking: Lotus Notes/Domino 5 [8/64]... (8xMPI) DONE
Raw:    3227K c/s real, 3227K c/s virtual

Benchmarking: Lotus Notes/Domino 6 More Secure Internet Password
[8/64]... DONE
Many salts:     90820 c/s real, 90820 c/s virtual
Only one salt:  53788 c/s real, 53788 c/s virtual

Benchmarking: Lotus Notes/Domino 6 More Secure Internet Password
[8/64]... (4xMPI) DONE
Many salts:     345597 c/s real, 345597 c/s virtual
Only one salt:  206224 c/s real, 206224 c/s virtual

Benchmarking: Lotus Notes/Domino 6 More Secure Internet Password
[8/64]... (8xMPI) DONE
Many salts:     677244 c/s real, 677244 c/s virtual
Only one salt:  402405 c/s real, 402405 c/s virtual

And the other tests I did yesterday, this time on 8 threads (i7 has 4
cores with HT)

Benchmarking: M$ Cache Hash 2 (DCC2) PBKDF2-HMAC-SHA-1 [128/128 AVX
intrinsics 8x]... (8xMPI) DONE
Raw:    4182 c/s real, 4182 c/s virtual

Benchmarking: M$ Cache Hash MD4 [32/64]... (8xMPI) DONE
Many salts:     116829K c/s real, 116829K c/s virtual
Only one salt:  47149K c/s real, 47149K c/s virtual

Benchmarking: MS SQL SHA-1 [128/128 AVX intrinsics 8x]... (8xMPI) DONE
Many salts:     79225K c/s real, 79225K c/s virtual
Only one salt:  49717K c/s real, 49717K c/s virtual

Benchmarking: NTLMv2 C/R MD4 HMAC-MD5 [32/64]... (8xMPI) DONE
Many salts:     5339K c/s real, 5339K c/s virtual
Only one salt:  4356K c/s real, 4356K c/s virtual


... as you see, mscash2 speed is about the same as on the CUDA. This
really surprised me here. Also, everyone's all-time favorite hashing
function, I like the pretty numbers:

Benchmarking: Raw MD5 [128/128 AVX intrinsics 12x]... (8xMPI) DONE
Raw:    148364K c/s real, 148364K c/s virtual


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.