Digifort UK has collaborated with Dell to enhance servers for both its VMS and neural analytics operation.
According to the company, VMS servers are focused on video processing and need CPU (Central Processing Units) for optimal operation, whereas analytics servers require GPU (Graphics Processing Units) for analytics processing.
Digifort is a Nvidia technology partner, designing its software for optimal performance on Nvidia GPUs.
With smaller systems, the GPU can fit within the VMS desktop server, reducing costs.
However, in the larger applications, the VMS and analytics servers are separate, with the analytics linked to the VMS camera channels.
Digifort UK opted not to present these as an OEM server range, choosing instead to run with the Dell brand and pass hardware savings on to customers, who also benefit from Dell’s next day warranty support, the company reports.
“Our VMS servers are typically 2U and come with 16, 24 or 32 Intel or AMD processing cores,” said Nick Bowden, MD, Digifort UK.
“They can control and record over 100 networked cameras each. RAID options for the operating system, OS (RAID 1), and storage (RAID 5) are standard, guaranteeing ‘uptime’; preventing data loss; and ensuring data protection compliance.
“Further resilience options include server failover, protecting against server or network outage.
“This performance is essential for high-security, mission-critical applications.
“A 7-year warranty is available, reducing the cost of ownership, with users alerted to server, OS, or storage drive faults in real-time, for preventative maintenance.
“Servers use less hardware per camera than NVRs, reducing power consumption, rack space and air conditioning requirements.
“Analytics server performance requires GPU, which is typically measured in CUDA cores. Our standard server builds easily deliver 3000 plus CUDA cores or more.
“As a technology partner of Nvidia, Digifort’s software is optimised for their GPUs.
“The processing ‘budget’ is distributed across all the analytics channels, with more demanding analytics’ rules receiving more of the performance.
“We claim over 30 neural analytics channels per 1U server, but this can be far more if simpler analytics’ rules and logic are deployed.”