hashcat 4.0.1 more slow than 3.6 (GT 740M)

[gunderhold]

Im using Nvidia GT 740M, so, using hashcat 3.4, i got 40000H/s with workload = 3. When I use hashcat 4.0.1, i got 10000H/s.... using a weak hardware it is hard, reducing your capabilities its terrible....

[Hydraze]

Hi, which hashmode is this? Are you using the -O flag with hashcat 4.x?
Starting with hashcat 4.x hashcat is using "pure" kernels by default. Those kernels accept longer candidates than optimized ones, at the cost of some "lower" performance.

[gunderhold]

Hi, which hashmode is this? Are you using the -O flag with hashcat 4.x?
Starting with hashcat 4.x hashcat is using "pure" kernels by default. Those kernels accept longer candidates than optimized ones, at the cost of some "lower" performance.

im cracking wpa/wpa2.

[atom]

Try -w 4

[asirena]

atom, you should use the same parameter in both cases in order to compare the results...

If you use -w4 with one of them and -w3 with the other one, you are rigging the results, aren't you?

Code:

- [ Workload Profiles ] -

  # | Performance | Runtime | Power Consumption | Desktop Impact
 ===+=============+=========+===================+=================
  1 | Low         |   2 ms  | Low               | Minimal
  2 | Default     |  12 ms  | Economic          | Noticeable
  3 | High        |  96 ms  | High              | Unresponsive
  4 | Nightmare   | 480 ms  | Insane            | Headless

gunderhold, please write here both commands you have used.

[royce]

atom, you should use the same parameter in both cases in order to compare the results...

If you use -w4 with one of them and -w3 with the other one, you are rigging the results, aren't you?

While I approve of making sure that we know exactly what the original poster's command line is, your approach here is not friendly.

No one is gaming any results. And it's a bit cheeky for you to show up out of the blue and start trying to teach *the main author of hashcat* about how to make fair comparisons.

So much has changed between 3.4 and 4.0 that the best course of action is to advise people how to get the best performance that they can.

[asirena]

royce, I think I have been polite, haven't I? And I haven't said any lie, have I? So I don't understand your words against me (I would like to ask you if you can avoid things like that in the future, please).


Well, there is a comparison between the speed of different versions. And it is a very interesting topic (at least in my opinion). Think about this: everybody want to crack their hashes the quickest as possible; so if there are other versions quicker than the last one, what is the problem to point that? No everyone will need the improves which the last version includes** but the speed always is welcome.

Return to the parameters. I cannot know what the original poster's command line is till he says it. In fact, he haven't answered whether he was using -O parameter yet. That is the reason I asked it.

Furthermore, obviously it is not the same to use a -w3 that -w4. So if the new version need a -w4 to get the same speed that older versions without that, really you must think there are any strange...

Because, in the best course of action you can use -w4 in the old versions too and again you'll get much better results, won't you?



** e.g. I need certain features which were removed (long time ago) from the tool. So I need to use a really old and slow (that does not even use the GPU!) version for that kind of tasks. Even I need to modify the system date to be able to use that. But there isn't another option...

[atom]

The meaning of -w 4 changed massively with 4.x, otherwise i would not have asked