Hashcat hashcat mask attack hafl md5 5100 slow speed

[davidchoi1992]

Hey everyone,

I’ve been running a mask attack on half-MD5 (Hash-Mode 5100) using 8× RTX 5090, but the performance seems much slower than I expected.

Here’s the speed I’m getting:
Session..........: hashcat
Hash.Mode........: 5100 (Half MD5)
Hash.Target......: f385c433805c6cd3
Time.Started.....: Wed Nov 26 14:44:41 2025 (26 secs)
Time.Estimated...: Wed Nov 26 14:46:45 2025 (1 min, 38 secs)
Kernel.Feature...: Optimized Kernel (password length 0-55 bytes)
Guess.Mask.......: CCC?1?1?1?1?1?1?1?1?1?1?1?1 [15]
Guess.Queue......: 1/1 (100.00%)
Speed.#01........:  8954.3 MH/s (1.39ms) @ Accel:160 Loops:8 Thr:1024 Vec:1
Speed.#02........:  8927.5 MH/s (1.41ms) @ Accel:160 Loops:8 Thr:1024 Vec:1
Speed.#03........:  8954.0 MH/s (1.39ms) @ Accel:160 Loops:8 Thr:1024 Vec:1
Speed.#04........:  8941.4 MH/s (1.40ms) @ Accel:160 Loops:8 Thr:1024 Vec:1
Speed.#05........:  8954.4 MH/s (1.39ms) @ Accel:160 Loops:8 Thr:1024 Vec:1
Speed.#06........:  8851.5 MH/s (1.40ms) @ Accel:160 Loops:8 Thr:1024 Vec:1
Speed.#07........:  8928.6 MH/s (1.40ms) @ Accel:160 Loops:8 Thr:1024 Vec:1
Speed.#08........:  8954.4 MH/s (1.38ms) @ Accel:160 Loops:8 Thr:1024 Vec:1
Speed.#*.........: 71466.2 MH/s
Recovered........: 0/1 (0.00%) Digests (total), 0/1 (0.00%) Digests (new)
Progress.........: 1856667648000/8916100448256 (20.82%)

Speed expected should be: 149.2 GH/s (19.01ms) @ Accel:16 Loops:1024 Thr:1024 Vec:1

For this amount of GPU power, the numbers feel low. Did I misconfigure something, or is this normal for half-MD5 in Hashcat? Any advice on tuning, recommended settings, or things I should double-check would be appreciated.

[penguinkeeper]

If you remove the "CCC" from the start, it should speed back up to normal. The reason for this strange behaviour is that Hashcat mostly uses the first 4 characters to parallelise from, so you may notice that CCC[mask] is drastically slower than [mask]CCC with it appended instead of prepended, this is just a flaw in Hashcat's internals that is unfortunately being hit quite hard here. You're not doing anything wrong

[davidchoi1992]

If you remove the "CCC" from the start, it should speed back up to normal. The reason for this strange behaviour is that Hashcat mostly uses the first 4 characters to parallelise from, so you may notice that CCC[mask] is drastically slower than [mask]CCC with it appended instead of prepended, this is just a flaw in Hashcat's internals that is unfortunately being hit quite hard here. You're not doing anything wrong

Thanks for the explanation that makes a lot more sense now. I didn’t realize the prefix affected Hashcat’s parallelisation like that., appreciate your help