Speed error

[dimler]

Hello. 
I have such a problem In my task the number of candidates is slightly more than 37 billion. The search speed is 1120 kH/s and it takes more than 6 days to complete 100% of this task. 

Question: h/s is not candidates per second? or do I not understand something correctly? Since at a speed of 1120 kH/s (if it is a candidate per second) I can sort through more than 96 billion passwords in a day. And in my case, it turns out that for 37 billion candidates, I need 6 days. That is, my actual brute force speed is about 71.3 candidates per second. 

Please explain to me what is my mistake and what I don't understand?

[s3in!c]

Where do you have these numbers from?: `I can sort through more than 96 billion passwords in a day.`

Bitcoin wallets are not a very fast algorithm, and also there may be differences between each wallet file as different hash iterations could be applied.

[dimler]

Where do you have these numbers from?: `I can sort through more than 96 billion passwords in a day.`

Bitcoin wallets are not a very fast algorithm, and also there may be differences between each wallet file as different hash iterations could be applied.

I counted based on the speed of 1120 kH/s*60*60*24=96 768 000 в сутки.
Or is the kH/s iteration rate not equal to the number of passwords per second??

[s3in!c]

Yes your calculation is correct, 1120kH/s means that it can test 1.12 million passwords per second.

The estimation in Hashtopolis is just calculated based on the completely finished chunks (so the 9% in your keyspace which are running are not accounted into the calculation at all), so it might be a bit off sometimes.

Do your agents' speed actually sum up to the total speed of 1120kH/s which is given? On the agents page check the speeds per agent to see if they are correct.

[dimler]

Yes your calculation is correct, 1120kH/s means that it can test 1.12 million passwords per second.

The estimation in Hashtopolis is just calculated based on the completely finished chunks (so the 9% in your keyspace which are running are not accounted into the calculation at all), so it might be a bit off sometimes.

Do your agents' speed actually sum up to the total speed of 1120kH/s which is given? On the agents page check the speeds per agent to see if they are correct.

I have a total of 24 video cards in operation, 1660Super and 2060, and in total they give out + - 1120kH/s.
But the problem is that 37 billion in 6 days is 15.5 times less than 1120kH/s, it's not just 9% less.

Here is the command I use: #HL# -a3 -w4 -D2 --force -1BCDFGHJKLMNPQRSVWXYZ -2bcdfghjklmnpqrstvwxyz -3CDGHJKLNPQRSVWXYZ -4aeiou ?3?1?1?2?4?2?2?1?4?1-

Maybe I'm doing something wrong, which is why there is such a big loss of speed?
Since if I just run Hashcat on one computer of 6 video cards, the speed there corresponds to the total time to complete 100% of the task, and through Hashtopolis there is a speed loss of about 15.5 times.

[s3in!c]

Why do the agents do not have a benchmark result value? There should be something else than just 0.

[dimler]

Why do the agents do not have a benchmark result value? There should be something else than just 0.

Because it was on this task in the settings of the task itself that I set a fixed number of tasks.
I'm experimenting with different settings to identify the cause of the speed loss.
Before that, the same task was with the default settings: (Use static chunking: - NO) but the result is the same and I need about 6 days for 37 billion passwords

[dimler]

Why do the agents do not have a benchmark result value? There should be something else than just 0.

Here is a screenshot of how the agents look on the standard settings.
But almost nothing changes since it takes 9 hours to sort through 37 million passwords at a speed of 1140 kH/s, and not 8 days, as indicated in the attached screenshot.
(8 days since the task has just been created, in 4-5 hours the number of days will drop to 6)

[s3in!c]

Are you running with rule files or is it a wordlist only?

[dimler]

Are you running with rule files or is it a wordlist only?

No, I don't use any rules or wordlist.
Here is my attack command: #HL# -a3 -w4 -D2 --force -1BCDFGHJKLMNPQRSVWXYZ -2bcdfghjklmnpqrstvwxyz -3CDGHJKLNPQRSVWXYZ-4aeiou ?3?1?1?2?4?2?2?1?4?1-
But I think I found the problem.
I ran the same command on hashcat and here's what it shows me (see the screenshot)
In the Restore.Point line shows exactly this value of 37 billion, and in the Progress line it already shows the total number of password combinations (629 billion).
That is, the total number of combinations is 629 billion, not 37 billion. But for some reason, in Hashtopolis, the values from "Restore" are displayed in the Keyspace size: line.Point" and not from the string "Progress".
I think it would be correct and much more convenient to display the total number of all combinations in the "Keyspace size:" line.