Hardware and Software Requirements
CHardware Requirements
Requirements differ based on how you plan to use the application, as the number of images, resolution, and modules used can significantly affect the amount of GPU memory required.
With GPU being the most important hardware component, strong CPU can significantly improve inference time and RAM with sufficient capacity is required.
GPU Requirements
To use Pekat, an NVIDIA GPU with CUDA compute capability 7.5 or newer (check on NVIDIA website) and at least 8 GB of memory (VRAM) is required (the memory consumption depends on used modules and image sizes).
For training AI modules we recommend at least NVIDIA GeForce® RTX 3060 (12 GB) or above.
We strongly recommend using desktop for both training and inference as desktop GPUs are usually more powerful and have better performance.
CPU Requirements
The most important factor regarding CPU when using Pekat is it’s Boost Frequency. We recommend using newer generations of Intel i5 or higher (i7 and so on) or equivalent AMD CPUs.
The minimal recommended boost frequency of the CPU is 4 GHz.
The impact of CPU is more apparent the bigger the feature size used for training is. Meaning that impact of CPU on training and inference is minimal when none of the modules used in project have feature size set above 32.
RAM requirements
The size of RAM again, depends on the use case, but the smallest should have at least 8 GB. However using RAM with 16GB or more is recommended.
For price-sensitive projects a less powerful NVIDIA GPU may be used, however, this can result in longer processing times. Additionally, GPU memory requirements may change if additional modules or higher image resolution are required.
For more information on GPU memory usage, please contact support at PEKAT VISION Customer Portal or contact support@pekatvision.com
Software Requirements
Operating system
Supported systems are mainly Windows 10 or Windows 11.
Other
NVIDIA GPU drivers are available here: