+- hashcat Forum (https://hashcat.net/forum)
+-- Forum: Developer (https://hashcat.net/forum/forum-39.html)
+--- Forum: hashcat (https://hashcat.net/forum/forum-40.html)
+--- Thread: How to generate hashes for directory content (files) not crack them ? (/thread-7226.html)
How to generate hashes for directory content (files) not crack them ? - reinis2006 - 01-23-2018
Hi !
I see hashcat is the one of rare tools using GPU horsepower to speedup cracking passwords but my intention is a bit different. I'm searching for tool that can use GPUs but for hash generation (SHA-256 - has type 1400 by https://hashcat.net/wiki/doku.php?id=example_hashes) not password cracking.
My use case would be scan a ton of archive files and compare the duplicates much faster using GPUs than CPUs.
Current benchmarking proves a i7 type CPU can't saturate more that 1-2GBit/s bandwidth on this task on any type of files (even 100GB each running parallel 16 threads) but testing hashcat on my Dell XPS 15 laptop with a week GT 750M is already 25% faster than the CPU and me having a local and network storage that is ~ 1GB/s (8Gbit/s) capable I would use the GPUs for the task as that would limit the need for so many workstations to do the work.
Looking at the docs and features I see hashcat is capable of multitasking and even multi node setups but what it would take to add this mode to the feature set in near future ?
Thanks !
Best regards,
RE: How to generate hashes for directory content (files) not crack them ? - undeath - 01-23-2018
imo this is unlikely to be implemented, but I'm not a hashcat dev.
also see this related thread: https://hashcat.net/forum/thread-7221.html
RE: How to generate hashes for directory content (files) not crack them ? - Chick3nman - 01-23-2018
In this case, GPUs will actually NOT be faster. Since the archives are very large, you will lose a lot of the things that make GPUs faster for password cracking when trying to load them in and calculate their hashes. Just look at the difference between the optimized speeds(-O) and the pure kernel speeds. For SHA256 the optimized kernel is limited to 55characters i believe, where the pure is out at 256characters for its limit. The drop in speed from a 55char pass to a 256char pass being supported is incredible. Now imagine a password that's 1billion characters long because its actually the binary data from an archive. Yeah, no. Far better off using CPUs imo.