Home > Products > Windows Passwords > Windows Password Recovery > GPU FAQ
Frequent questions on GPU pasword recovery
OneNote password recovery v3.4
Support for Microsoft Office 365
Reset Windows Password v14.2
Telegram data recovery, Photo Database and Media Player investigation tools, and some more
Office password recovery tools
Resetting VBA passwords
New blog post
Dumping the history of users' IP addresses in Windows

Articles and video

You may find it helpful to read our articles on Windows security and password recovery examples. Video section contains a number of movies about our programs in action

Windows Password Recovery - GPU FAQ

Q: What are the system requirements for the program?
A: The program supports NVidia video cards with CUDA compute capability 3.0, AMD Radeon 7xxx, Intel 4xxx, and higher GPUs. The full list of CUDA supported devices can be found here. Compatible AMD Radeon cards are shown at the Wikipedia site. Besides, you should also have the latest video drivers installed.
Q: What versions of Windows the program supports?
A: Password recovery using GPU acceleration is supported starting up with Windows XP (NVidia GPUs) and Windows Vista (AMD GPUs) on both 32-bit and 64-bit systems.
Q: How do I know which architecture does my video card support?
A: For NVidia devices:
Launch the program, open the menu 'Options - General Options,' select the 'GPU Settings' tab, select 'NVidia CUDA' platform, and choose your video card here. The 'Compute capability' field in the description section should display your GPU architecture.
For AMD devices:
Launch the program, open the menu 'Options - General Options,' select the 'GPU Settings' tab, select 'AMD OpenCL' platform, and choose your video card here. The 'CL_DEVICE_VERSION' and 'CL_DEVICE_OPENCL_C_VERSION' fields should display your GPU architecture supported.
Q: Where can I get the latest video drivers?
A: You can download the latest drivers from NVidia and AMD websites.
Q: Where can I read more info about CUDA?
A: Wikipedia site is a good starting point to start from.
Q: Where can I read more info about AMD/ATI Radeon cards?
A: http://en.wikipedia.org/wiki/Comparison_of_AMD_graphics_processing_units
Q: Where can I read more info about Intel Graphics?
A: https://en.wikipedia.org/wiki/Intel_HD_and_Iris_Graphics
Q: After I launch a GPU-based attack, my computer freezes or crashes into BSOD. What’s the problem?
A: The problem may be caused by the following reasons:
  • Your video card had been overclocked, and it was malfunctioning at high load. If that’s the case, bring the frequencies of the video memory/cores to its defaults.
  • Insufficient or ineffective cooling of your card. When you launch a GPU-based attack, the program utilizes most of the GPU power, and the GPU temperature rises to a critical level. Make sure that your video card is well cooled, the GPU slot and your system unit are free from dirt and dust. Unwise use of some video settings may have a negative impact on the video card’s temperature and its stability under high load conditions. For example, some applications reduce the fan speed to minimize the noise, which does result in noise reduction but also increases the core temperature.
  • Power supply problem. Your card can consume a lot of energy at full load, and the power supply unit may be unable to handle such a high demand for power. If the video card has additional 6-pin or 8-pin power connectors, make sure they are all properly connected. 
Q: When I launch a GPU attack, my computer slows down a great deal. How can I fix that?
A: By default, the application is set up for using video cards of medium performance. That’s usually 256 threads per block, 256 blocks, and 1000 passwords per thread. For older video cards such a configuration is too much and may cause a slowdown. Consider reducing the value of 'Passwords per thread' to 100 or even less.
Q: What's the best way to find optimal values of 'Thread blocks' and 'Passwords per thread' in the GPU attack settings?
A: You can do that either empirically or by doing some maths. For example, if the values are 100 and 100, and the average speed of attack is 1 billion passwords per second, you can calculate that the GPU kernel is called about 390 times per second (the number of passwords calculated each time is usually 256 * ThreadBlocks * PasswordsPerThread). Naturally, the fewer calls, the less the overhead, and the higher the attack speed. On the other hand, you must call the GPU program at least a couple of times per second. So use a calculator, and adjust the parameters. You can also adjust them using a rule of thumb, that is, increasing their values until the speed of attack stops going up and the computer slows down. If you have a GPU monitor installed in your system, it should indicate a load of at least 98-99 percent. Besides, it’s important to know some other things too. First, don’t set the summary values of those parameters too high. Otherwise, your system may malfunction or freeze. Second, you’d better not set the value of 'Passwords per thread' at less than 100 as it will negatively affect the speed of attack regardless of what kind of video card is used.
Q: Does the PCI-Express bus have any impact on the performance?
A: Actually, this impact is negligible. It’s usually masked by other factors. So the generation of your PCI-Express bus and its performance don’t matter much.
Q: Does the amount of video memory matter?
A: No, it doesn't. However, in most cases, your GPU should have at least 256 Mb of video memory.
Q: What's the maximal number of GPU devices does your program support?
A: It depends on your hardware. Even though the program supports up to 255 devices, typically, up to 8 devices can be installed into a 4 PCI-E slot motherboard (4 double-GPU cards).
Q: A GPU-based attack slows down my PC so I can barely use it. How can I fix it?
A: There are two ways to fix it: temporary and permanent. As a temporary fix to the problem, go to the attack settings and try reducing the number of GPU blocks used or the number of passwords checked per GPU thread. As a permanent fix, install a second video device, provided that you have a second slot on your motherboard, and that your power supply unit can handle the additional load. For example, you can use some cheap card as the primary one (for displaying information on your monitor), and a second, more powerful one, for brute-forcing passwords.
Q: I have more than video cards in my computer. Can I use them all for brute-forcing?
A: Yes. You can use all or some of them. Just open general settings and specify the GPU device(s) to be used by the program.
Q: Can I brute-force passwords on devices which performance varies a lot? Say, GT8600 and GTX580?
A: Yes, you can.
Q: Your application can’t use all of my GPUs.
A: You will have to disable SLI in order to be able to use all devices.
Q: The program cannot detect my video card. What should I do?
A: Update your video drivers. If it didn't help, try to extend your desktop to all devices (if you have more than one device). Re-plug your device into another PCI-Express slot.
Q: Can I use both NVidia and ATI devices simultaneously?
A: Yes, you can use NVidia, AMD, and Intel devices simultaneously.
Q: How can I check my GPU utilization?
A: Open Hardware Monitor. In 'What to show' drop-box choose the device you need and select 'Show' to display it. You can then click 'Start' or 'Stop' buttons to manage the hardware monitoring. The GPU monitor shows device load (utilization), temperature, and fan speed.
Q: My NVidia GPU is absent in the hardware monitor?
A: You should install/reinstall the NVAPI library. Download the library at https://developer.nvidia.com/nvapi
Q: My AMD GPU shows zeros in hardware monitor.
A: Install/reinstall the latest AMD drivers or ADL components. Make sure your AMD device is active (connected to the active monitor). Non-active devices are not processed properly by ADL due to a bug in AMD drivers.