Userful just announced the release of Userful v8.5 which brings some powerful new features for video walls. In particular the 8.5 release features a pivotal new breakthrough in performance and scalability for the Video Wall by adding a single Nvidia GTX cards-- We call them Nvidia Offloading Cards —into the host PC. This extra horsepower brings very significant improvements to:
- Scalability — support for 60+ displays, instead of the 25 for the previous version.
- Image Quality — switching from linear scaling to cubic / lanczos scaling to substantially improve visual image quality on heavily scaled video walls.
- Performance — scales up to 6K source content resolution vs 4k currently. Also, offloading 50%+ of the workload to the Nvidia GPU frees up the CPU resources.
- Synchronization — substantial synchronization improvements.
The conceptual leap built into Userful’s breakthrough approach to video walls is that each display receive (in real-time over the network) only their portion of the processed video, instead of having displays directly attached to the server. This essentially means that the video wall can grow to any size and configuration you want limited only by the the host server’s processing capability and available network bandwidth.
Userful uses commodity PC hardware, but by adding Nvidia’s GTX line of consumer graphics cards into the mix as an Offloading Card we can significantly increase the processing capacity of the host PC, and with 10Gig network cards and 10Gig uplink ports on switches now becoming relatively inexpensive, we can soar past any previous CPU and network limits on Userful’s scalability. The key difference between the Userful video wall (which communicates with displays over the network) and traditional video wall controllers where displays are physically connected to a traditional video cards, is scalability. With traditional video wall controllers you have a fixed number of cards you can physically install, which limit the number of displays you can deploy in your video wall. No such limits exist with the Userful solution.
What does this mean for system integrators?
We are recommending everyone take advantage of the offloading card capabilities. This means that Userful resellers may need to choose a PC chassis that supports one and possibly two nvidia expansion cards. Nvidia CUDA cards are available in both full and half height, however it is much easier and less limiting to standardize upon full-height cards (hence a full size chassis) with a very powerful power supply (e.g., 850 watts).
To test out the Offloading Card feature you would need to add in some additional hardware, namely:
- We recommend the GTX 960-series Nvidia GPUs as being the best cost/performance. But if you need a half-height card you can instead use the less expensive GTX 750 series GPU and still free up significant CPU resources.
- A Gigabit switch is still adequate for most deployments, but for the larger deployments you can consider establishing a 10-gig uplink between host and switch so you would then need a 10Gig NIC [$350] and a 10Gig uplink switch [$800-1500] in addition to the host PC.