Your Browser is Out of Date

Nytro.ai uses technology that works best in other browsers.
For a full experience use one of the browsers below
Computer Vision Workflow Enhancement – Image Restoration with ProHawk
Overview Solution overview Partner overview Architecture Sizing and scaling guidance
Sizing the solution
Scaling the solution
Use cases Conclusions and References

Sizing the solution

Home > Workload Solutions > Computer Vision > White Papers > Computer Vision Workflow Enhancement – Image Restoration with ProHawk > Sizing the solution

Sizing the solution

  • The ProHawk sizing is important because if the VM is underpowered then the real-time restoration of video will be slowed down potentially delaying the detection of a critical event.

    ProHawk Base Requirements

    The base requirements (per stream) as stated by ProHawk are:

    • 4 vCPU (2.2 GHz)
    • 500 CUDA cores
    • 500 MB Memory per stream
    • 5 Mb ingress
    • 15 Mb Egress

    It is important to note that hardware-based video decoding and encoding needs to be supported by the GPU. A large portion of the time spent by ProHawk servers relates to decoding then reencoding video streams.

    Base VM Size

    After extensive testing with both the A16 and the A40 the optimal VM size for ProHawk is as follows:

    CPU
    8
    Memory
    16 GB
    GPU
    1/4 of A16 GPU (equivalent to A2)

    This base VM can support a total of 2 ProHawk streams. The details relating to optimal sizing are in the use cases section.

    Milestone XProtect Smart Client Plugin

    If ProHawk video restoration is integrated into the Milestone XProtect Smart Client for playback, then it is critical to ensure that GPU resources are made available to the Smart Client. When designing a system, it is important to identify how many Milestone XProtect Smart Client plug-in installations will be made as this will reduce the number of GPUs available for ProHawk real-time processing.