News for hackers. Stuff that matters.
Re:Relaunched TrueCrypt site with original content and docum... by kijin on Wednesday August 12, 2015 @07:47AM (#10047195)
10 seconds is ridiculous, especially if you know what you're doing and use strong passphrases.

The passphrase to one of my TrueCrypt volumes is around 30 characters long. It's not completely random, so let's say it only has 2 bits of entropy per character. That's 60 bits total. A brute-force attack would take an average of 2^59 guesses before it succeeds.

At 10 seconds per guess, we're looking at approximately 180 billion years to crack my passphrase. Even if you devoted a million computers to the job (as an extremely well-funded adversary might do), it would still take 180,000 years.

If you weakened the PBKDF to take 1 second instead, it would take 18,000 years to crack my relatively weak passphrase using a million computers. That's still long enough that I don't care at all. Moreover, if I were really paranoid, I can easily bring it back to the 180,000-year mark (or more) by adding a few more characters to my passphrase, which would only take another second to type. So why wait 10 seconds?

Re:Relaunched TrueCrypt site with original content and docum... by monort on Wednesday August 12, 2015 @08:25AM (#10047309)
You can bruteforce much faster on GPU.

E.g. 20*10^9 ripe160 hashes per second on 8x AMD R9 290X (price is around $3000): https://hashcat.net/oclhashcat/

60 bits are not secure against attacker with medium budget.

Re:Relaunched TrueCrypt site with original content and docum... by kijin on Wednesday August 12, 2015 @09:07AM (#10047472)
Then we can use GPUs to speed up VeraCrypt's PBKDF, too.

The point is that there's no need for a PBKDF that takes 10 seconds on modern hardware.

Sure, that 60-bit passphrase is weak. But it takes a trivial amount of effort to increase the entropy of a passphrase. Let's say I increased it to 80 bits. In terms of brute-force resistance, that's equivalent to increasing the PBKDF's iteration count 1 million times.

Re:Relaunched TrueCrypt site with original content and docum... by luck87 on Wednesday August 12, 2015 @08:36AM (#10047341)
Some years before oclHashCat support, I buil a gpu cracker for truecrypt volume TrueCrack (https://code.google.com/p/truecrack/) . To be honest oclHashCat has better performance. Now the board of the oclHashCat tests ( AMD R9 290X ) costs about 300 euro (not $3000).
Re:Relaunched TrueCrypt site with original content and docum... by monort on Wednesday August 12, 2015 @09:17AM (#10047511)
They used 8 cards for that test.