When working with PL-X 10 GigE cameras, there are some additional considerations to help ensure optimal performance from the camera. These "tips, tricks and got'chas" regarding NIC selection, settings, cables, and computer configuration are listed below.
Network Interface Card (NIC):
Selection and placement:
Choose a high quality / high performance PCI-E 3.0 (or better) NIC with at least x4 bus interface, x8 (or better) is recommended for optimal performance. We like the Marvell (Aquantia) AQtion 10 GigE NIC.
Be sure to plug the card into a PCI-E slot which is capable of using all (4 or 8 or more) PCIe channels.
Configuration:
NIC Settings:
The camera supports oversized, or ‘Jumbo’ Ethernet packets, and the use of Jumbo packets is necessary to use full network bandwidth. For most NICs, Jumbo packets are turned off by default.
To achieve high levels of performance, edit the NICs configuration to enable Jumbo packets to at least 8192 bytes (set it to the maximum setting). To enable Jumbo packets in Windows, right click on the NIC from device manager, select properties, and the option to enable Jumbo Packet will be available from one of the tabs.
Also, if configurable, the amount of receive buffer allocated to the NIC should be maximized. Note: This setting may not be configurable on all NICs.
Finally, be sure the NIC is operating in full duplex, 10 Gigabit Ethernet mode. This should be done via the default auto-negotiation.
IP Address:
The camera must be on the same IP network as the NIC (as defined by the IP address and subnet mask). The Pixelink cameras support both DHCP (requires a DHCP server) and LLA to automatically choose IP addresses. However, if LLA is being used, it’s important to recognise that it can take upwards of 20 seconds for both the NIC and the camera, to choose appropriate IP addresses. Furthermore, the user has the option of assigning the camera a persistent (non-volatile) IP address if they prefer to manage the IP information themselves.
If the camera is not on the same IP network as the NIC, then most Pixelink applications (applications built using the Pixelink API) will not be able to locate the camera. However, the tools FindGevCameras and the Pixelink IP address Tool will still see the camera, and the latter will allow the user to re-configure the IP address information of the camera.
Windows Firewall:
Pixelink applications should setup so that they are not ‘blocked’ by the Windows Firewall. That is, all Pixelink applications should be added as ‘apps allowed to communicate through the Windows firewall’. The Pixelink installer will add the necessary permissions to accomplish this for the Pixelink Capture (P-Cap), Capture OEM (C-OEM), and the Firmware Updater application, but you may need to add exceptions for other camera applications.
The first time any application (built using the Pixelink API) is run, Windows will detect that it is attempting to communicate to a device through the firewall, and will display a "Windows Security Alert" prompt.
When prompted, it’s important to check to allow communication over all three network types, as in the screenshot below. If you do not do this from the pop up, it can be added later via the Windows Firewall program.
If the application is not permitted through the Windows Firewall, the application will still be able to communicate with the camera, but only at lower performance levels; it will not be permitted to use the ‘Jumbo packets’ referenced above.
Computer configuration:
Streaming at or near the full 10 Gb/s places some very stringent real-time demands on the system. Although Windows was not designed to be a real-time system one can improve real-time (like) characteristics by doing things like:
Close non-critical applications, especially ones that may consume large amounts of memory
Be cognitive of applications that may be running in the background, things such as virus checker/scanner programs
Minimize disk activity. This includes avoid swapping applications that may result in disk activity related to page swapping by the Windows memory manger
Connecting the camera:
Use high quality Cat 6a Ethernet cabling, respecting Ethernet cable length rules.
Furthermore, be aware that adding media converters and connection adapters, add further attenuation which may reduce useable cable lengths.
If the camera is connected via one or more switches, ensure that these switches have sufficient buffering to operate at full wire speed, and that they too are configured to support Jumbo packets.
Notes on maximum cable lengths:
Car6a (recommended): 100m maximum
Cat6a with PoE: 60m maximum
Cat5e: 30m maximum
For all other cable types, refer to the 10GBase-T specification for compatibility and recommended lengths.