3

Earlier I added bcrypt with cost 12 (15 is a bit too long for login, 12 looks about right).

I'm also planning to add the https://howsecureismypassword.net/ open source version but I'd rather not scare people signing up.

The default example has "calculationsPerSecond: 10 billion" -- is there any information on how many password attempts could be done against bcrypt with cost 10-14 on Today's hardware?

The most recent information I found is an Ars Technica 2012 article describing seemingly 71k/sec with better than average hardware: 25-GPU cluster cracks every standard Windows password in <6 hours -- it doesn't mention the cost used though

3

From 2014, hardware included

I've found a paper which looks like it contains the facts you need.

There is a table which shows the performance of certain devices corresponding to the cost factor:

Cost / Device   12        10        8         5
Epiphany 16     9.64 c/s  38.7 c/s  151.3 c/s 1207 c/s
Zynq-7020       64.83 c/s 253.1 c/s 932.6 c/s 4571 c/s
Zynq-7045       226.3 c/s 888.6 c/s 3371 c/s  20538 c/s
HD 7970         35.76 c/s 142.9 c/s 569.2 c/s 4556 c/s
FX-8120         42.93 c/s 171.2 c/s 680.2 c/s 5275 c/s
Xeon Phi 5110P  50.18 c/s 200.7 c/s 800.8 c/s 6285 c/s
i7-4770K        53.67 c/s 214.2 c/s 852.8 c/s 6615 c/s

It is worth noting that three devices, Epiphany 16, Zynq-7020 and Zynq-7045 are low-power parallel platforms. You can read more about this in the paper I linked at heading numbers 3, 3.1 and 3.2

From June 2016, hardware not mentioned.

This website shows the times to hash a password with bcrypt (and some other hashing functions) depending on the cost.

This image from the website shows the timings for bcrypt with cost values ranging from 6 - 20.

Hashes per second

  • Cost of 10: 14.7 hashes per second
  • Cost of 11: 7.3 hashes per second
  • Cost of 12: 4.4 hashes per second
  • Cost of 13: 1.9 hashes per second
  • Cost of 14: 0.9 hashes per second
  • That's just his own machine though? From what I can see it doesn't mention specs...could be an old P4 – Dird Sep 25 '17 at 21:56
  • I've edited my answer to include another source which has hardware specifications included. – Joe Sep 26 '17 at 7:23
  • Thanks :o c/s = attempts per second? I'll assume 100m calcs/sec is way too high then in the present day – Dird Sep 26 '17 at 10:19
1

Maybe this table can help? It gives hashrates on 8 NVidia GTX 1080 GPUs.

The comments mention the following:

sroussey commented on 21 Oct 2016 I understand the desire to keep bcrypt at work factor 5, but can you do additional tests with a factor of 10 and 15. Best to start the 15 now and have a history of how it progresses. And 10 is common today. Thanks!

epixoip commented on 6 Dec 2016 @sroussey the work factor is log2, so you can simply divide the hashrate in half any number of times to get the hash rate for higher work factors. For example, 105700 H/s for cost 5... 105700 / 2^(10-5) =~ 3303 H/s for cost=10, 105700 / 2^(15-5) =~ 103 H/s for cost=15. No further benchmarking necessary ;) But the reason why we benchmark with static parameters like this is so we can easily compare device performance spanning several hardware generations.

So it seems that you can calculate the hashrate on 8 GTX 1080 GPUs for all parameters of bcrypt. For cost 12 it would be 105700 / 2ˆ(12-5) = 825 hashes per second. For 1 GPU you could divide this by 8 (rough estimate).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.