09-25-2019, 07:07 PM
I wonder what is the fastest (or at least a fast) way to generate millions of hashes like MD5 and the SHA family (224, 256, 512)? Do you guys know any scripts/tools that will read lines from files, hash each line while taking advantage of GPU, especially CUDA?
A fast solution would be to just read lines with cat and then pipe each line to sha1sum as an example, but maybe there is a faster way to generate hashes using GPU? As far as I know with little CUDA knowledge, we need to read lines from a file with the CPU and copy them to a GPU device and then we can calculate the hash. Maybe there is a way to copy all lines read in one step to the device, which is may be faster. While using CPU only, the copy step is ommited, so maybe the generation process could be faster when we use CPU power instead of GPU power? What do you think?
Unfortunately I can't do a performance analysis on my own since I think there is no option to abuse hashcat as a hash generator (*) and I was not able to find tools that does this exactly.
*really sad, would be a very nice option. I know that hashcat is open source but I don't think it is a one line adjustment in the code to get this option, right?
A fast solution would be to just read lines with cat and then pipe each line to sha1sum as an example, but maybe there is a faster way to generate hashes using GPU? As far as I know with little CUDA knowledge, we need to read lines from a file with the CPU and copy them to a GPU device and then we can calculate the hash. Maybe there is a way to copy all lines read in one step to the device, which is may be faster. While using CPU only, the copy step is ommited, so maybe the generation process could be faster when we use CPU power instead of GPU power? What do you think?
Unfortunately I can't do a performance analysis on my own since I think there is no option to abuse hashcat as a hash generator (*) and I was not able to find tools that does this exactly.
*really sad, would be a very nice option. I know that hashcat is open source but I don't think it is a one line adjustment in the code to get this option, right?