Crack/Clock

time-to-crack, on real GPUs

Type a password and see how long three NVIDIA V100 GPUs would need to brute force it, benchmarked with Hashcat on the KAUST Ibex cluster. The choice of storage algorithm changes the answer by a factor of millions.

Your password never leaves this page. Everything runs in your browser. No network requests, no storage, no logging.

Password

Waiting
This is one of the most common passwords in the world. It is in every attacker's wordlist and falls instantly, whatever the math says.
Time to crack if stored as NTLM, a fast legacy hash
—

Attacker hardware

power 739 GH/s on NTLM

If stored as

AlgorithmCracking speedTime to crack

Full keyspace brute force, an upper bound. Real attacks try wordlists and rules first, so predictable passwords fall faster.

How this works

Character set size comes from the classes you use (lowercase 26, uppercase 26, digits 10, symbols about 33). Keyspace is charset to the power of length, average guesses is half of that, and time is guesses divided by the measured GPU speed, scaled by the number of GPUs.

Measured baseline, 3 Tesla V100, Hashcat 7.1.2

Fast hashes in the billions per second: NTLM 246, MD5 166, SHA-1 61, SHA-256 21, SHA-512 7.3, SHA3 5.0 (GH/s). Real KDFs far slower: PBKDF2-SHA512 3.1 MH/s, sha512crypt 1.0 MH/s, bcrypt 236 kH/s, Argon2 3,760 H/s. Fast hashes have no work factor, so they give no real protection no matter how strong the name sounds.

About

Academic cybersecurity project. Not an official KAUST service and not affiliated with or endorsed by KAUST. No live cluster is contacted; all figures are constants.