I've been using the old cudahashcat 2.x until I reinstalled my system. I realized I cann't find any newer release on cudahascat so I assume it is included in hashcat but when I look at their github it seems to be only the OpenCL is included. So my question being do I need to find cudahashcat mod somewhere or should I just use the official hashcat 4.x?
3 Answers
No, neither the hashcat binaries download nor by hashcat download via Kali Linux repositories include the Nvidia CUDA Toolkit; the Nvidia CUDA Toolkit must be downloaded separately- feel free to review my full gist on GitHub. Hope that helps!
Installing the Nvidia CUDA Toolkit 10.0
Installing the Nvidia CUDA Toolkit 10.0 for Ubuntu 18.04 - Option A (Recommended)
sudo apt update && apt dist-upgrade -y && reboot
lspci -v
sudo apt install -y ocl-icd-libopencl1 nvidia-driver nvidia-cuda-toolkit
.nvidia-smi
- Run
sudo hashcat -I
orsudo hashcat --benchmark
for benchmarks.
Installing the Nvidia CUDA Toolkit 10.0 for Ubuntu 18.04 - Option B
- Go to Nvidia CUDA Toolkit 10.0 for Ubuntu 18.04.
- Select the following Select Target Platform Window: Linux > X86_64 > Ubuntu > 18.04 > runfile (local)
- To check your OS architecture run
uname -a
it should return "x_86_64" within the terminal display.
- To check your OS architecture run
- Download the Base Installer (2.0 GB) by click on the Download button.
- Access Terminal and close X Server to avoid errors while updating nVidia driver by running
sudo init 3
; a black-screen will appear and prompt you to enter your Ubuntu username and password- please enter. - After logging-in, run
cd ~/Downloads
. - Run
sudo sh cuda_10.0.130_410.48_linux.run
- Scroll down to accept the conditions by holding down the "Return" key or space-bar- upon scrolling to the bottom, type:
accept
; alternatively, you can hit 'CTRL + D' to bypass scrolling. - Install the CUDA 10.0 Toolkit by enter
y
- Do you want to Install the OpenGL Libraries
y
- Do you want to run nvidia-xconfig? Enter
y
- Install the CUDA Toolkit? Enter
y
- Unless otherwise preferred, hit the "Return" key to enter the specified default absolute-path (e.g.
/usr/local/cuda-10.0
). - Enter
y
to install a symbolic link at/usr/local/cuda
. - Enter
y
to Install the CUDA 10.0 Samples - Hit the "Return" key to enter CUDA Samples default location-path (e.g.
/home/alexanderjsingleton
). - After the install, open the X Server by running
sudo init 5
; if for whatever reason you're unable to reboot the computer after installing the CUDA 10.0 Toolkit, turn-off the computer and reboot- remember to click on the Settings icon and select "Ubuntu on wayland" before entering the username and password. - Access terminal and enter the following commands in sequence:
sudo add-apt-repository ppa:graphics-drivers/ppa
sudo apt-get update
sudo apt-get install mesa-common-dev
sudo apt-get install freeglut3-dev
- Reboot the computer by running
sudo shutdown -r now
You should definitely just use the latest hashcat from hashcat.net. Note that CUDA is not included and downloaded separately.
There have been many changes and improvements since 'cudaHashcat' / 'oclHashcat' / 'hashcat' (CPU) were all merged into the new unified 'hashcat'.
All devices now require OpenCL - even CPU - and CUDA is completely gone.
[Update Jan 2020: CUDA support is returning in the upcoming 6.0.0 release of hashcat. One major benefit: you will have access to all available GPU memory - not just 1/4 of it! The CUDA driver itself is still a separate download from NVIDIA.]
-
What happens if I have an older GPU/CPU that doesn't really run OpenCL (well)? I have a GF116 [GeForce GTS 450 Rev. 2] in this particular machine. Nov 23, 2017 at 2:56
-
If the card is still supported by the manufacturer (current drivers), and OpenCL works generally, then hashcat should work. Nov 23, 2017 at 4:09
-
-
Ah, bummer. Only cards supported by current drivers are supported by hashcat. (It's a little OT, but you can get a much newer card quite cheaply used - even a GTX750Ti would be significantly better, a GTX 970 should also be relatively affordable.) Nov 24, 2017 at 2:40
-
1Yep, CUDA support is returning in the upcoming 6.0.0 release of hashcat! If you can't wait, you can grab the beta (hashcat.net/beta) and/or the latest GitHub (github.com/hashcat/hashcat) Jan 24, 2020 at 15:18
Since hashcat v6.0.0, CUDA seems to be supported. See https://hashcat.net/forum/thread-9303.html.