Decreasing speed with an increasing number of GPUs - Printable Version +- hashcat Forum (https://hashcat.net/forum) +-- Forum: Support (https://hashcat.net/forum/forum-3.html) +--- Forum: hashcat (https://hashcat.net/forum/forum-45.html) +--- Thread: Decreasing speed with an increasing number of GPUs (/thread-12427.html) |
Decreasing speed with an increasing number of GPUs - whitemerc - 12-23-2024 Here is cl: hashcat -m 23800 -a 3 -o cracked_hash.txt hash.txt n?1?1?1?1?1?1?1?1 --increment --increment-min=8 --increment-max=12 -1 abcdefghijklmnop -O -S -d 1,2,3,4,5 only 1 GPU 3070 Speed.#1.........: 21415 H/s (25.48ms) @ Accel:1 Loops:16384 Thr:512 Vec:1 2 GPUs (3070 + 3060Ti) Speed.#1.........: 15372 H/s (25.38ms) @ Accel:2 Loops:16384 Thr:256 Vec:1 Speed.#2.........: 13892 H/s (20.69ms) @ Accel:1 Loops:16384 Thr:512 Vec:1 Speed.#*.........: 29265 H/s 5 GPUs (2 3070 + 3 3060Ti) Speed.#1.........: 5805 H/s (25.45ms) @ Accel:1 Loops:16384 Thr:512 Vec:1 Speed.#2.........: 5556 H/s (21.10ms) @ Accel:1 Loops:16384 Thr:512 Vec:1 Speed.#3.........: 5854 H/s (21.14ms) @ Accel:1 Loops:16384 Thr:512 Vec:1 Speed.#4.........: 5895 H/s (25.24ms) @ Accel:1 Loops:16384 Thr:512 Vec:1 Speed.#5.........: 5542 H/s (21.48ms) @ Accel:1 Loops:16384 Thr:512 Vec:1 Speed.#*.........: 28652 H/s So 5 GPUs are slowly than 2 GPUs core i7 6700, 48G RAM, ssd. CUDA.Version.: 12.2 ubuntu 22.04 hashcat 6.2.6 Any suggestion how to fix this? RE: Decreasing speed with an increasing number of GPUs - penguinkeeper - 12-23-2024 See: https://hashcat.net/wiki/doku.php?id=frequently_asked_questions#how_to_create_more_work_for_full_speed RE: Decreasing speed with an increasing number of GPUs - whitemerc - 12-23-2024 After studying the forum and documentation, I realized that my CPU and bus speeds are insufficient to assign tasks to the GPUs. Currently, when running hashcat, the CPU is at 100% usage, and many hashcat processes are waiting in the queue with 0% load. Upgrading my CPU and motherboard even to the 11900KF will not solve this problem. I see the only way to use maskprocessor... Will try |