Help with gpu

[Crypttoad301]

I have two instances running each with 8 h100 gpus one thou is doing way better cant figure out why
How could i change the accell to match the quicker one

[TheBogeyman]

I have two instances running each with 8 h100 gpus one thou is doing way better cant figure out why
How could i change the accell to match the quicker one

What is hash you are trying to crack,, I mean hash mode

[Crypttoad301]

I have two instances running each with 8 h100 gpus one thou is doing way better cant figure out why
How could i change the accell to match the quicker one

What is hash you are trying to crack,, I mean hash mode

11300 on both same wallet file on each

[Crypttoad301]

What im trying to figure out is how the accell is so much greater in the first application

[TheBogeyman]

What im trying to figure out is how the accell is so much greater in the first application

I think that you can be happy with the speed for this hash type. In my opinion it is quite high,.. If you run benchamrk hashcat -b -m 11300 -O how much speed do you get?

[Crypttoad301]

What im trying to figure out is how the accell is so much greater in the first application

I think that you can be happy with the speed for this hash type. In my opinion it is quite high,.. If you run benchamrk hashcat -b -m 11300 -O how much speed do you get?

I dont understand why the second instance configured same way is only half of that

[slyexe]

I dont understand why the second instance configured same way is only half of that

You are apply more work to one attack than the other. In the first instance you have a wordlist with rules, in the second instance you are strictly using a wordlist. 

Check out the FAQ
https://hashcat.net/wiki/doku.php?id=fre...full_speed

[Crypttoad301]

But the one with rules shouldnt be faster

[Crypttoad301]

And same exact thing on other instance is 290 kh on lther its 790 kh

[TheBogeyman]

Wordlist which is aprox 85GB large with rule behaves differently than dictionary only with aprox 170GB. Also you have all on risers, bandwidth might be a problem. 
Just speculating, or maybe alfo hashcat decides wordlist to chunks and first iretirates chunk with rulle and if not found, sends another chunk to GK. 
With wordlist only, chunks are split. Maybe wordlist is faster but riser might be problem. 

At this point devs might help to answer.