Hardware

Graphics card cracks passwords 25 times faster

Russian company Elcomsoft has released a new version of its flagship Elcomsoft Distributed Password Recovery software that now leverages on GPU hardware acceleration to crack Windows NTLM passwords up to 25 times faster than previously possible using the highest-end desktop PC.

Russian company Elcomsoft has released a new version of its flagship Elcomsoft Distributed Password Recovery software that leverages on GPU hardware acceleration to crack Windows NTLM passwords up to 25 times faster than previously possible using the highest-end desktop PC.

Using a GeForce 8 graphics card via Nvidia's CUDA framework, cracking an eight-character Windows password now takes only three to five days, instead of two months. It is expected that multiple computers can be combined in a cluster to increase the throughput even further.

Excerpt from heise Security:

The Compute Unified Device Architecture (CUDA) framework allows high-performance tasks to be outsourced to the graphics card. In particular, tasks that can be highly parallelized are especially suitable for modern graphics cards like Nvidia's, with its 128 stream processors. GPUs do not handle floating-point calculations as accurately as CPUs, which is why they are not as suitable for such applications as climate modelling, but they are adept at the high precision fixed-point arithmetic that is used in encryption.

You can check out Elcomsoft's press release here. (pdf)

Elcomsoft has announced that it will be incorporating this patent-pending technology into its entire family of enterprise password recovery applications, using up to four separate video cards that are supported by high-end PC motherboards.

Folks, we're not even talking about quantum computing yet. Is there really such a thing as "unbreakable" encryption?

--------------------------------------------------------------------------------

Stay on top of the latest tech news

Get this news story and many more by subscribing to our free IT News Digest newsletter, delivered each weekday. Automatically sign up today!

About

Paul Mah is a writer and blogger who lives in Singapore, where he has worked for a number of years in various capacities within the IT industry. Paul enjoys tinkering with tech gadgets, smartphones, and networking devices.

73 comments
dvijaydev46
dvijaydev46

If an 8800GT takes 3-5 days, a Radeon 5770 doesn't even take 19 hours, the GPU cracks the password 3-6 times faster.

JK
JK

Why would I use this when a Rainbow Table can crack it in form 18 minutes to two hours and only costs $499! JK@for-sec.com

tomhirtler
tomhirtler

Many users have something like [ProperName+number] for a password. That hardly requires a brute force attack and most of them are the minimum 8 characters. Until we get that hole plugged the rest of the argument is merely amusing nonsense.

JohnMcGrew
JohnMcGrew

...when programmers bypass the code-bloat that is Windows. When 9 out of 10 clock cycles is lost to OS overhead, it's a wonder we get anything usefull done.

P K Pal
P K Pal

Passwords can be cracked, however, it still depends on the hardware (for speed) and software(s) (for encryption). Now how about this...... Encrypt the password using one algorithm. The resultant password can be re-encrypt using another algorithm and so on with the 3rd encryption. This can be done with atleast 3 different algorithms to produce a password that will be difficult to decrypt unless one knows the reverse steps exactly. This is possible only if the file to be passworded allows such an encryption that needs to be built into the system (and is kept out of preview of the prying eyes).

OldER Mycroft
OldER Mycroft

If it WERE a Windows password it would not be limited to 8 characters. I also find it odd that no-one has given any thought to the old favourite (from way back in DOS days) of Failsafe as a deterrent. I used this a number of years ago [where and why is not up for discussion] and its beauty was that the 'form' of typing was also included in the password. The design strategem was that even if the password was cracked, you still needed the user to enter it, since the timing of the keystrokes could be metered by the program and the identity of the 'inputter' brought into the formula. A user under stress would be of no use to someone attempting to break-in, since the presence of higher than normal levels of adrenalin would cause the user to enter the password in an unacceptable (to the program) sequence of keystrokes. That system was harder to break than any encrypted password formula, it was thought.

Tig2
Tig2

I use a multiple pass encryption with a few twists that I would love to chat about but won't. I don't think it can be broken. When we think about encryption, we have to consider the tools that can break it- and then think past them. One time pad is unbreakable. Without the pad and a knowledge of the specific daily algorithm, you aren't going to get through, no matter what you try. There are other approaches that are as secure. But they all depend on the devious nature of the programmer that creates them. And the ability of that programmer to keep their mouth shut. Even in the shift to quantum computing, a devious mind can code in a manner to beat it. We can teach a computer to use brute force. The real question is if we can teach it finesse. Until we do, MY personal encryption methods are safe.

apotheon
apotheon

That's why I prefer one-way hashes with salts: rainbow tables aren't so useful in those circumstances.

w2ktechman
w2ktechman

Actually, many companies have adopted stricter policies over the last 5 years in regards to PWD's. And many have also done a lot for educating users about PWD's. For instance, I worked at 1 place that required 12 characters, alphanumeric, etc.. My current place requires 8 min, alphanumeric, Upper and lower case, special characters, cannot be words from a dictionary, and cannot be recognizeable names.

Absolutely
Absolutely

But do those users have anything worth cracking?

Absolutely
Absolutely

When 95 out of 100 users cannot use the machine [b]at all[/b] without the GUI you're calling "overhead", how much of that is really code-"bloat"?

steven.alker
steven.alker

We?ve been sponsoring some basic research on quantum computing. There are two threads, one is the massive parallel capabilities of a true quantum computer ? but this is only going to be available at some time in the mid-future. That would facilitate enormous cracking capabilities of standard and double-pass keys. The other is the variable quantum-key where the states of two keys are entangled. One can be varied by an algorithm or even randomly and the key required at the ?other? end must match the state of the first quantum key. At the moment, the limitation is 4 bits, so realistically; a hacker can skip through all combinations in whatever window is presented for the decryption process. With 32 bits and a short window, this would be much more difficult. With 128 bits, it would be down to luck if a hacker could find the right combination at any given time. The challenge is to make the quantum pairs robust enough to reliably hold 128 bits of information over a reasonable physical distance. That will occur a lot sooner (I think!) than the massive parallel quantum computers which can be used to crack conventional ciphers ? including Dan Brown?s fictitious ?revolving cipher? If such a thing existed, the massively parallel computer could crack it within a ?revolution? of the cycle! Steve Alker

bikingbill
bikingbill

Didn't the Wehrmacht believe their Enigma codes were unbreakable?

Dumphrey
Dumphrey

are only unbreakable if the encryption used is free of flaws in both design and implementation. Many times, cracking encrypted files has little to do with passwords, and more to do with finding the flaws in the build which carry over into th encrypted file. That being said, one time pads are the overall strongest encryption.

jonscott8
jonscott8

Thanks for the laugh, I needed it this morning.

mhbowman
mhbowman

If it can be put together, it can be taken apart. Whether it's free satellite, computer access, or using a bump key to open a sophisticated lock there is ALWAYS a way around security. It doesn't matter if the reward is untold millions or bragging rights, people don't like being told they can't, or that it isn't possible.

Absolutely
Absolutely

[i]Actually, many companies have adopted stricter policies over the last 5 years in regards to PWD's. And many have also done a lot for educating users about PWD's.[/i] Yes, but those same users still go home and login with [catName0] or [MMYYYYDD], if they use passwords at all. DDoS attacks tend to hit companies, from home users, according to the vast majority of analyses I've read. So, companies should just blacklist all commercial ISP's IP ranges! Now, back to work.

JCitizen
JCitizen

probably a password generator would have been more practical. But no body had anything worth stealing, it was just to comply with HIPPA. Unless you think stealing the social security number of an indigent person would make any money. Maybe the password generator would have a vulnerablility? Can't win for losing!

shardeth-15902278
shardeth-15902278

It seems people often forget that in addition to looking at likelihood of being hacked, one should also look at the value of the protected content. Putting a $2000 car alarm in a $200 car is quite pointless.

JohnMcGrew
JohnMcGrew

Well, for starters you've got registry residue, countless DLLs from long abandoned code still running, countless security patches checking over the doings of other countless security patches, crapware that nodbody knows is still there...

JCitizen
JCitizen

I would think 3d video graphics would help implement my idea; because to generate a true chaotic algorithm you'd need at least three vectors. An x, y, and z axis. The key would be the attractant which would probably take eons of super computing power to crack especially if you added more vectors. But of course if you owned a botnet you would have a huge super computing power base. Am I sounding crazy? Criticism is welcome. :p

Mr L
Mr L

...it was thought to be practically unbreakable. i.e. unbreakable by any practical method and therefore safe to use at that point in time. Turns out that wasn't the case; and that innovative mathematics, machinery, and some luck can overcome a lot. If your point is that nothing is unbreakable, I think most people will agree from a theoretical perspective. From a practical standpoint, I believe there are schemes that are "safe enough" for most data or identity protection needs.

Tig2
Tig2

You are speaking to the thing that breaks encryption more assuredly than any other thing. Failure to account for the flaws. In good encryption we have to consider the first possible port of compromise- the end user. If the end user is compromised, so is the password, and thus the system, regardless of safeguard. Second port of compromise- failure on the part of the programmer to exhaustively test both their logic and their code. As you say, the flaws are simply doors. Finally, the third port- arrogance. Oddly I have been accused of this very thing in this thread. But oddly, you still know nothing of the thinking that led to my encryption program. Therein a critical key. Get me to talk about the thinking that supports the code and you will know how to decrypt my files. But if I don't talk, you have no shortcut. When done properly, a one time pad is virtually unbreakable.

Tig2
Tig2

Until you prove me wrong. But in order to do so, you will need to get over the assumptions that you made when you read my post. I haven't a clue what they are. Believe me, we are thinking in brute force attacks. What happens when finesse is a requirement? Ask the tough questions. It does pay off.

Tig2
Tig2

I think that the human mind is capable of twists and turns that a straight line logical pathed vehicle is not. Therefore, I believe that it is still possible to develop encryption that cannot be broken. So I'm an optimist. :)

xmlmagician
xmlmagician

I will second that...in addition to that really good hackers they go around passwords. Windows have so many wholes that you can get what you want without troubling your self with passwords

shardeth-15902278
shardeth-15902278

The vast majority of people use the same password for everything. Yahoo, myspace, work... SO when they do get socially engineered, it doesn't really matter how complex the password was. It could be a 256 character monstrosity, it still only took 2 minutes to acquire it.

JCitizen
JCitizen

for the amount of work put into it anyway! :D BTW - What did you think of my spintronics idea?

Digicruiser
Digicruiser

Now that is a brilliant idea, you are going to be the manager of this company? I find it funny though that there are password managers out there and it takes one password to reveal it all. I still like to know from other contributors here, how well do their "clients" cope with 12 character password!!!! They remember it if it's easy but do they write it down first when they change it so they can drum into their heads??? I think someone said Stenography scrambles the picture and is unencrypted - not quite - you have the same picture but you mix in an encrypted passworded text or file and you can't tell visually that it's happened. You can test for a payload but the rest is up to the encryption ability of the product that encrypted it in the first place. I use this occasionally at work because even though it's against the law for this department to read employee's E-mails (they need a warrant etc), we still have people in EDS randomly snooping our mail, so anything sensitive has be done another way - they have no hope breaking stenography with better stenography software around. Lastly, I can start your business immediatley, I would be able to collect all the pieces of paper where they write credit card numbers, passwords to documents, sensitive stuff which in short time they forget about (having the piece of paper) - plenty business here! Cheers

HAL 9000
HAL 9000

When I worked Bank Security I could buy a CD with thousands of Card details for $25.00 US and then set loosly to my current area as well. I still remember allowing the bank to pay for one of these CD's and sitting down in the Chief Of Security Office and using the phone to bill him for a 6K Sound system. He was really unimpressed when I used his name & card details to pay for the purchase and even more horrified when the sale was accepted. :D Naturally the bank did [b]Absolutely[/b] nothing about that issue and continued to blame the customers for every case of Credit Card Details getting loose because it was easier. Col

steven.alker
steven.alker

How about setting up a crime syndicate to steal all the little bits of paper that people write their passwords on? They will just think that they've lost them. Then we sell to some Wally who thinks that there is value in the crap which most people encrypt. Did you know that the going rate for a VISA card number, name and address has dropped to $0.50 (According to "The Daily Telegraph")? Doesn't look like we'd get much for our trivia, does it? Steve Alker

mhbowman
mhbowman

When you have password generators you'll see an increase of post-it's with the password written down. Like the T-shirt says: "Social Engineering: Because There Is No Patch To Human Stupidity" Left to their own devices most user passwords are their car model, pet's name, kid's name followed by a number: current year, kid's age, or numbers 1-5 used in succession. Example: Honda1 - Honda5, and the series starts over again. For that reason we're currently looking at single sign on, in conjunction with a badge reader. When you remove your card, the computer logs you out.

Absolutely
Absolutely

If an alarm is expensive but easy to defeat, it will tend to attract car thieves. I'd be surprised if there are not a few analogous indicators among crackers.

Absolutely
Absolutely

I think the market rewards cheap & fast, for a variety of reasons. One, it seems to work -- until a vulnerability is exploited before it's discovered by a White Hat, then deemed worthy of patch by another White Hat in the corporation(s) having legal access to the affected source code, then rewritten & deployed. Two, on top of the compulsory ignorance of the working of particular programs that I just pilloried, there is the voluntary ignorance that's so funny in Help Desk threads. Anyway, the market doesn't exactly demand crap, it just doesn't demand functionality very eloquently.

JohnMcGrew
JohnMcGrew

...now + cheap. There's a motto I tell my clients: "Good, Cheap, Fast. Pick two". The critical mass of consumers has not yet come to understand what quality is. Well, perhaps they do. We could buy Macs, but then again we're too cheap.

Absolutely
Absolutely

The market generates them via economic demand for apps that work now, with security testing maybe later -- by the thief community, if at all.

JohnMcGrew
JohnMcGrew

...I think it comes down to the unfortunate realities of development cycles in the software biz. Programmers are pressured to get things "working" under what they pervceive as tight schedules. They do so, under the belief that after they get something "working", that they will be able to go back later and then clean it up. But the reality is that managers, under pressure to work on the "next thing" decide to "leave well enough alone" and nobody ever gets to go back and clean up. That is, at least until some sort of problem pops up, at which time it's a total crisis and something needs to be kludged asap; forget about elegance. The result? The multi-gigabyte mess that is Windows.

Absolutely
Absolutely

I refer you to the outcry that results whenever Microsoft obsoletes any technology. Victims of their own success. It does all come down to end-user convenience, starting with but not limited to the GUI.

JCitizen
JCitizen

We need more input from contributors like you. Of course someone might think you are spreading complete misinformation; as how would most people know? It has been a while since I studied it and I had forgotten about the strange attractor "factor". No Longfellow pun intended. :D I had always read, and thought I understood, that knowing the starting point or "attractant" was precisely what made chaotic systems so unpredictable. Especially with more than three vectors. But if I understand you correctly in a natural system where nature holds the keys, it is an exceedingly difficult system to predict. However if a man holds the key any other man can always then predict the attractant condition because that would be decipherable. In nature there is always a different starting point and hidden vector, to add to the complexity. To mimic this you attempt to camouflage this with a graphic key? Would adding a hidden vector make unauthorized decryption any more difficult? I don't know; but it seems like the rat race to come up with better encryption follows closely with processor speed and longer prime numbers. It just seemed like there had to be a better way. Perhaps spintronics would be a temporary answer. In a spintronic system each data set has an infinite condition for each data point. Since the spin of an electron can't be predicted but apparently controlled, you could build a spintronic hardware device that uses more than 0's and 1's. An eight bit data string could theoretically hold an infinite number of spin attitudes. Even the controlling laser could change light frequency to add confusion to the uninitiated black hat. The article in Scientific American led me to believe the best use of this technology was discovered accidentally when lab workers noticed a "pooling" affect on some silicon wafer material they were analyzing with a laser beam. Subsequent investigation discovered that they could change the spin attitude of the electrons in the material and actually control it well enough to predict a super mass storage idea. With each data point capable of holding a huge number of semipermanent attitudes you could get the knowledge of the universe on an inch squared piece of substrate! To hold the key? If I am misinformed about this as well; I am incapable of understanding clear text about physics and should shoot myself immediately! :p

steven.alker
steven.alker

No, you are not crazy but perhaps you are a little misinformed. The odd properties of a chaotic system or a strange attractor are that a tiny error in the start condition when entering the attractor or chaotic equation, will lead you to an unpredictable error in your vector coordinates after so many iterations or passes. It doesn?t matter how many dimensions you use ? two are handy for visualisation, three for visualising vector spaces and as many as you like if you can handle the maths. The point is that applying a strange attractor to a fixed number, with a fixed number of iterations of the system, will always leads you to the same result. All you need to know is the algorithm for the attractor. If you don?t know that you can use standard decryption to test the key, which would presumably be the vector coordinates of the start position against the exit condition (That would be the users key) As this relationship never changes, its only a matter of time before you pop then answer out of interrogating the key. The other area of chaos maths is that any line drawn through a representation of a strange attractor will return a string of numbers (If it is interpreted as numbers ? it could be colour a la Mandelbrot Poster or it could be ASCII code) Those numbers from one coordinate in a given vector space to another are unique ? but unless you know where to look, with infinite accuracy, you?ll never find them. Buried in the ASCII decoding of a transit of a Lorenz Attractor are the complete works of William Shakespeare, evidence that in infinite number of monkeys were not involved in writing them, this posting and the code for an as-yet unbreakable encryption, followed shortly by the solution of how to break the unbreakable encryption. If only you knew where to look and at what resolution! Steve

JCitizen
JCitizen

before a machine was captured, using one of the first (if not the first), electronic computers; or electro mechanical, I can't remember if it used vacuum tubes or not.. http://www.bletchleypark.org.uk/ The capture of an actual machine happened soon after, or at the same time; and that confirmed their reverse engineered copy. With an exact copy of the code wheels they could crack the code wheel setting in a minute with the computer, and set the code wheels in minutes. The fact that some submariners were late changing the code wheels wouldn't have mattered. They would have only missed a few lines of messaging..

sboverie
sboverie

The story I heard about enigma was that the system was used by the German navy and army. One of the services got lazy and did not reset the base code as often as the other branch of service. This allowed multiple messages using the same code setup and that helped the allies break the code. Having the enigma machine helped but it was the human element that made the break through.

ijusth
ijusth

deleted duplicate posting sorry

ijusth
ijusth

dleted duplicate posting sorry

ijusth
ijusth

plus they got a copy of the physical device - something the Germans couldn't plan for. U481 (or whatever then number LOL) was based on that although the British did the hero stuff and not the Americans (oops just a SLIGHT change of history hehe)

Tig2
Tig2

I appreciate that you took the time to read what I was saying and respond to it. I think that the place where a complex level of encryption has the most value is data at rest intended to remain at rest. I agree that the approach is akin to swatting a fly with a sledgehammer- somewhat overkill. I know someone who does investigation as an essential job element. That person uses the encryption method I outlined but does so because the data gathered may become evidence in a court. In that scenario, it becomes easier to prove that the electronic evidence hasn't been tampered with- the key is stored separately from the encrypted data and can be held by a different person. I agree that the law of diminishing returns pertains as well. If I use strong password rules and a sufficient length string, I probably have sufficient protection. But I can see the day coming that won't be enough. As cracking methods get more sophisticated, I think that the best way to stay in front is to begin thinking out of the box as we establish an encryption methodology. I may not need it today, but I will bet that I will need it soon enough.

steven.alker
steven.alker

I believe from my own experience and engaging with our own specialists that you are essentially right. The caveat is protection that it would work for a given user, usually the inventor, but for a number of reasons might be unsuitable for mass use. I can protect everything of mine, in my office or on our network from our general workforce without any sophisticated encryption? from the Non-exec chairman through to the store man. It simply isn?t worth their time to bust an 8 letter password, the rewards are not great and the risk, of being caught is great. For more sensitive stuff, I rely on 128 bit encryption and it?s as safe as it is valuable. If someone wants to spend more money on cracking open my projections than the market value of those projections to my competition, then that?s fine by me. As my business is forecasting, I?m the only person who knows why they mean what they mean. In any case, I?ve always been sort of flattered that people will pay to spy on my guesswork! If something needed to be really secure, say because big-bad agencies were after it, I might use your multi-layer approach with finess, but don?t think that I?d bother. They?d either torture the key out of me (If I knew it) or kidnap my children. I?d tell them anything then. Steve

Tig2
Tig2

I can't help that you are only reading the words that you want to see. As I have said before, the key is not to add layers of complexity- 128 bit is better than 64 bit- but to add finesse. In WWII, the US developed a code that the Japanese were totally unable to break. For three years- 1942 through 1945, Japan tried and failed to decipher the code. What the US did was enlist the assistance of native Navajo tribesmen. The root Navajo language is difficult to begin with and nuances of words are literally not expressed vocally but with entirely different words. Further it is considered to be a "mother tongue"- it is not a written language but one that is handed down through generations. The Navajo Talkers were credited with being instrumental in the US winning Japan. Only one of many articles on the Navajo Talkers: http://www.history.navy.mil/faqs/faq61-2.htm My claim is that finesse will defeat a brute force attack. Why do I think so? Think about it. I have a document. By a simple letter swap- preferably by a single use pad- I now have an encrypted document. Now I encrypt it again using an obscure algorithm. Now I encrypt it a third time using a different approach. In order to de-crypt the document, you have to know what the series of steps were to encrypt it. Then you have to perform those steps in reverse order to get to the document encrypted by hand. THEN you have to have access to the one time pad that was used to encrypt the original. Hope that I didn't translate the original into an obscure language first. I don't think that it is arrogance to say that we don't tend to think of finesse when we consider cryptography. And assuming that I use the approach I have just outlined, I don't think that it is arrogance to say that you are not likely to get through to the encrypted document. And frankly, I had hoped that someone would actually come to the table and say, "Hey let me try!". That would have provided me with an opportunity to truly test the theory. And if that person WAS able to get through, to be able to improve the approach.

Timbo Zimbabwe
Timbo Zimbabwe

"Until you prove me wrong." So we are wrong unless we can prove a negative? OK, give us your IP address and we'll prove you wrong. Arrogance... pure and simple.

apotheon
apotheon

This is about brute force password cracking, not about cracking encryption. If you tell me to guess a number between one and ten, and I get up to ten chances, I'll get it right before I run out of chances. Increase the number of possibilities to twenty and tell me I only have time for fifteen guesses, though, and there's a chance I might fail. That's the real strength of a password system against a brute force methodology for cracking the password: increase the number of options. The security cracker's time for guessing is measured in the tension between CPU clock cycles and how old the guy's getting while he waits for his brute force attack to come up with an answer. As I pointed out in [url=http://blogs.techrepublic.com.com/security/?p=342][b]A little more about passwords[/b][/url] today, a twelve character password containing both upper- and lower-case letters, numbers, special characters on the number keys, and spaces turns that three-day eight-byte 52-possible-characters (for 52 ** 8 possible combinations) brute force attempt into a 3.6 million year twelve-byte 73-possible-characters (for 73 ** 12 possible combinations) brute force attempt.

Editor's Picks