Our software uses all the available computational resources efficiently: multiple CPUs, NVIDIA, AMD and Intel Arc GPUs and multiple computers (supported with Passware Kit Business, Forensic, Ultimate and Mobile).
GPU (Graphics Processing Unit) cards help accelerate password recovery by up to 400 times, compared to CPU-only systems. To assess your system capabilities for solving password recovery issues, use Hardware Benchmark to measure hardware performance on typical password recovery tasks (over a dozen of the most popular file types, such as MS Office, APFS, iTunes backup, RAR, BitLocker, etc.).
Passware supports all types of NVIDIA, AMD and Intel Arc GPUs with OpenCL v2.0 and higher. Mac version supports OpenCL acceleration on AMD GPUs and NVIDIA/AMD/Intel Arc eGPU.
NOTE: Hardware acceleration on Mac is an experimental feature dependent on Apple’s OpenCL drives, which are not 100% reliable.
Up to 12 GPUs are supported per host computer (depending on the motherboard). Some older NVIDIA cards might not work, please check our Troubleshooting section for more information.
Hardware acceleration is available for the file if there is a flag "Hardware acceleration possible" in the "Complexity" line:
The green "Yes" in the "GPU Support" column in the supported file types list points to the hardware acceleration support.
The password calculation process depends to some extent on the processor speed. We recommend at least 1 GB RAM for each GPU card.
We have tested Passware Kit on NVIDIA RTX 4090, AMD Radeon RX 6900 XT and Intel Arc B580 to provide you with the recovery speed for several file types. Please check it below:
https://www.passware.com/kit-forensic/performance/
Speeds for various GPUs and CPUs shared by our customers can be found on our Hardware Benchmark page.
Cost-efficient hardware
We recommend using Intel Core i7 processor or similar. The number of cores is more important than its frequency. There is a limit of 32 physical cores. Please also note that for each GPU Passware Kit uses 1 additional core that is equal to 2 threads.
We do not recommend anything specific for the motherboard and case, but pay special attention to the cooling system: it should provide a stable GPU temperature during the entire password calculating process. Passware Kit shows the temperature for the GPU cards in the interface and allows setting a max temperature or regulate the GPU load to avoid overheating.
To accelerate the password recovery process, use NVIDIA GeForce RTX, AMD Radeon RX or Intel Arc cards. The efficiency of password recovery depends on the number of CUDA/Steam processor cores and Base Clock. Lite Hash Rate (LHR) limitations should not affect Passware Kit performance.
AMD cards are cheaper than NVIDIA, providing the same or even higher performance. If you use GPU acceleration, pay attention to the corresponding cooling system and power supply unit, depending on the number of GPU cards. For hardware acceleration on some strong file types, it is recommended to have 2 x (RAM + GPUs RAM) of free disk space for the paging file.
Maximum performance
Maximum performance can be achieved by using Distributed Password Recovery. The more Passware Kit Agents you use, the better.
- Recommendations for Passware Kit Server:
Intel Core i7 or AMD Ryzen 5000 series processor or higher. No GPU. Disable the built-in Passware Kit Agent. - Recommendations for Passware Kit Agent:
Intel Core i9 or AMD Ryzen 9000 series/Ryzen Threadripper PRO 7000 processor, with 12 cores or more. NVIDIA RTX 5080 or AMD Radeon RX 7950 XTX (or higher) for hardware acceleration. Corresponding cooling system. Corresponding power supply unit.
Comments
4 comments
Greetings!
Would you go for a Quatro RTX 8000 card, or several RTX2080 ti cards?
Hello Harald,
The speed depends on the amount of CUDA cores, Quadro RTX 8000 has 4608 and RTX 2080Ti has 4352 cores. We would recommend to purchase several 2080Ti cards instead of one Quadro RTX 8000
Hi, do you have an updated benchmark list of GPUs for 2024?
Hello Ah,
Please contact our Customer Support for this information.
Please sign in to leave a comment.