+- hashcat Forum (https://hashcat.net/forum)
+-- Forum: Misc (https://hashcat.net/forum/forum-15.html)
+--- Forum: Hardware (https://hashcat.net/forum/forum-13.html)
+--- Thread: Building a cracking password infrastructure (/thread-3862.html)
Building a cracking password infrastructure - israellesg - 11-27-2014
Hello everyone,
I have been looking into building a nice infrastracture for discovering the password of around 24'000 files (excel, word, pdfs, pgp keys...).The budget that I have been given is of 50'000 euros. With that in mind I was looking at the most powerfull graphic cards currently in the market and I found the followings:
- GTX 980
- R9 295X2
- TESLA K40 (too expensive)
I would like to use around 8 of one of these. However, the problems that I find is that I don't know which one of these 3 is better and I what chassis with enough power supply and space to use.
What can you recommend me?
Thanks in advance for your help.
RE: Building a cracking password infrastructure - epixoip - 11-27-2014
GTX 980 is a great card, but R9 295X2 is not ideal for password cracking, especially not in a large multi-GPU environment. If you want to go with AMD then go with the R9 290X. Telsas are worthless for password cracking, so avoid them.
If you know where to look, there is a company that sell exactly what you are looking for pre-built & pre-configured with warranty & support.
RE: Building a cracking password infrastructure - Rolf - 11-27-2014
If your deadline is flexible, you could wait.
NV promised to release a high-end Maxwell Q4 2014 (seems more like Christmas sales 2014).
Some sources say it will beat GTX 980 in performance-per-watt category.
RE: Building a cracking password infrastructure - israellesg - 11-28-2014
Thanks for your answers. I will probably need to order the equipment this week.
Is there a reason why the GTX 980 is better for password cracking than the R9 295X2? Looks like the AMD card has more cores (double, althought they are slightly slower) and more memory.
The idea was to set up 6 computers with 2 R9 295X2 each working together with passware. The reasons was because of the heat and the power supply requirement. However, if you can explain me why GTX 980 is better I could just switch.
Another quick question, is it better to divide the graphic cards in several computers or to group the together in one as much as possible?
Thanks a lot for your answers, I really appreciate it.
RE: Building a cracking password infrastructure - epixoip - 11-28-2014
R9 295X is not ideal because it uses a hybrid cooler and has a very low Powertune max target temperature, which causes it to aggressively throttle. Dual GPU cards are almost never the way to go, but the R9 295X is just the epitome of bad dual-GPU cards. It is most certainly better to go with 2x R9 290X instead of one R9 295X.
GTX 980 has the best Perf/$ ratio on the market today. It is comparable to the R9 280X in single hash brute force performance, but faster than the R9 290X in multi-hash performance and other attack modes such as combinator attacks, at about half the power and heat of the R9 290X.
Building a password cracking cluster is not near as simple as "throw some GPUs into a chassis and start cracking passwords." There are a lot of design considerations, especially with heat and power concerns. Six nodes with four GPUs each will draw a lot of power and generate a ton of heat. For a 24-GPU cluster you really need to use special hardware that is built for this purpose, otherwise you will likely experience a ton of failures and headaches.
Also, why do you want to use Passware?
RE: Building a cracking password infrastructure - israellesg - 11-28-2014
(11-28-2014, 09:38 AM)epixoip Wrote: R9 295X is not ideal because it uses a hybrid cooler and has a very low Powertune max target temperature, which causes it to aggressively throttle. Dual GPU cards are almost never the way to go, but the R9 295X is just the epitome of bad dual-GPU cards. It is most certainly better to go with 2x R9 290X instead of one R9 295X.
GTX 980 has the best Perf/$ ratio on the market today. It is comparable to the R9 280X in single hash brute force performance, but faster than the R9 290X in multi-hash performance and other attack modes such as combinator attacks, at about half the power and heat of the R9 290X.
Building a password cracking cluster is not near as simple as "throw some GPUs into a chassis and start cracking passwords." There are a lot of design considerations, especially with heat and power concerns. Six nodes with four GPUs each will draw a lot of power and generate a ton of heat. For a 24-GPU cluster you really need to use special hardware that is built for this purpose, otherwise you will likely experience a ton of failures and headaches.
Also, why do you want to use Passware?
I see your point. I guess we will go for gtx980 then. Our plan is to run mainly a dictionary attack and the some brute forcing as the number of files to crack is quite big and we only have 180 days for accomplish this.
The client already has bought a Passware license prior we arrived. Is there any sustantial benefit using OclHashcat? If you could tell m the difference I would love to use hashcat with ubuntu serve.
Sorry for my unkownledge.
RE: Building a cracking password infrastructure - epixoip - 11-28-2014
For the algorithms that both Passware and oclHashcat support, oclHashcat is much much faster. Also, if you have a large number of files to crack, Passware does not support cracking them simultaneously. If you queue up multiple files, it will run each attack on each file serially!
By the way I sent you a private message, did you see it?