08-20-2017, 03:15 PM
The main problem with your approach (e.g something like "I just want to use the GPU to generate a hash list faster") is that the main bottleneck there for sure is I/O.
Even if you use the fastest SSD or ramdisk, all operations involved with I/O (all those instructions, caching/buffering etc) will take way longer and will undo any speedup you achieve by using GPUs.
That's also why hashcat doesn't show/print every (password) candidate it is currently hashing.
So for short, you shouldn't forget that the slowest part in the chain (the so called bottleneck) will determine how fast the overall process is. It doesn't make sense to use the fastest hardware for hashing (or let's better say to convert passwords to hashes and write them somewhere) if you know that there are some other operations (like writing to a file/disk/RAM etc) that are much slower and will reduce the speed by a significant amount.
Even if you use the fastest SSD or ramdisk, all operations involved with I/O (all those instructions, caching/buffering etc) will take way longer and will undo any speedup you achieve by using GPUs.
That's also why hashcat doesn't show/print every (password) candidate it is currently hashing.
So for short, you shouldn't forget that the slowest part in the chain (the so called bottleneck) will determine how fast the overall process is. It doesn't make sense to use the fastest hardware for hashing (or let's better say to convert passwords to hashes and write them somewhere) if you know that there are some other operations (like writing to a file/disk/RAM etc) that are much slower and will reduce the speed by a significant amount.