Tech Tip: Leveraging GPU Support to Display More Cameras in Command Client

Question: I need to display 40 HD cameras from a single Command Client workstation, but doing so overloads my computer’s CPU. This negatively impacts how the video is rendering on my monitors. Is there anything I can do to improve video decoding in Command Client?

Answer: Video decoding – especially the decoding of high-definition (HD) video with high frame rates – can put a lot of strain on your computer’s central processing unit (CPU).

Even a mid-level PC^[1] that’s supporting the display of 40 HD cameras (20 cameras each over two monitors) at 15 frames per second (fps) results in a CPU load of around 80%. This type of load would certainly impact workstation performance.

One way to avoid this strain on your CPU when displaying video is to take advantage of the hardware accelerated video decoding feature in March Networks Command. This feature leverages the support of commercial off-the-shelf Graphics Processing Units (GPU) for H.264 video decoding.

GPUs are popular devices today among gamers because they support image processing, allowing video game enthusiasts to play very vivid, lifelike games without impacting computer performance. People mining cryptocurrencies like bitcoin also use GPUs to support the mathematical calculations required to release these currencies.

But GPUs can also help with video surveillance display.

Starting with Command Client Release 2.5, you can use a GPU to shift the processing power off your workstation’s CPU, enabling the smoother display of live and recorded video and the ability to display more HD cameras on your Command Client workstation. Reducing your CPU load also frees up your computer to run other applications that may be necessary to your organization.

The benefits of accelerated video decoding are proportional to the quality of the video card you use, so the more powerful the GPU, the greater processing power you can harness in Command Client.

Using a GPU^[2] in the test case scenario outlined above (with 40 HD cameras) cut the CPU load in half, from 80% to just 40%.

The following chart illustrates some of our test results, with the X-axis showing the number of camera streams and the Y-axis showing the CPU load.

In testing with both 4MP and 2MP cameras at 15fps, using a GPU with Command Client cut the CPU load by an average of 40%.

With Command Client, you can actually add multiple GPUs to your workstation to display video on up to four monitors, as you would in a typical control room application. Provided that each GPU has at least one monitor attached, Command Client will leverage all the available GPUs.

Currently, this feature is available in Command Client, but will soon be available for all March Networks Command products.

Getting started

To get started in Command Client, you’ll need to purchase a recent GPU that’s compatible with Command Client. We recommend NVIDIA (for top performance, try NVIDIA GeForce GTX 1050 Ti or GTX 1080 Ti.)

1. Insert the GPU into your computer.

2. In Command Client, click the main menu button in the top left corner.

3.From the main menu, select Preferences.

4. Under the Preferences tab, there are sub-tabs. Select the System sub-tab.

5. Under Video Settings, click the box for “Enable Hardware Accelerated Video Decoding”.

6. Click Save to save the setting.

After enabling this feature, you should notice a substantial improvement in your video display.

Hardware accelerated video decoding is a great option for small and mid-size organizations that want to improve video display without having to purchase new computers. For large organizations that require an enterprise-level solution for video display, March Networks Command Center is a more ideal choice.

March Networks Command Center is a powerful video wall solution that can decode up to 32 streams per monitor. It scales to your specific requirements, so you can just add new monitors as needed. It also offers convenient alarm management features, like alarm process guidance and prioritization, for real-time incident assessment and resolution.

^[1] In this test case, we used a PC with an Intel® Xeon® E5-1630 CPU

^[2] NVIDIA Quadro M2000 graphics board.