Responses to user support requests. Weekly overview. Part 22.
Q: We want to be able to use the current graphics card for video processing and that nl relapses the load on the CPU, we have the AMD RX 9070 graphics which apparently is not compatible with the version we currently use Xeoma (22.8.16), we require to use this graphics card, and if it is a new version is it possible to give us the update for free or we must pay?
My other question is, if we use the T1000 graphics card can we use it in xeoma version 22.8.16 to process video and analytics? If so how can we know approximately how many cameras or analytics it supports.
A: This depends primarily on which side needs to be handling the decoding (which side actually has the GPU installed): server or client.
Server:
The ability to use AMD Radeon GPUs on the server’s side for hardware acceleration of decoding was added in Xeoma 24.2.14. The server version of Xeoma you can use is defined by your license’s update period, you can confirm it via Main menu -> Information -> Active licenses.
If the current license does not allow you to update to 24.2.14 or higher – yes, you will need to purchase a renewal for your license. Please note that It’s possible to get a one-time update, which is cheaper than a 1-year update (in case you are not interested in other updates).
Client:
The ability to use AMD Radeon GPUs on the client’s side was added to Xeoma around the end of 2023. The client does not require a license to unction, so it’s possible to keep an older version of Xeoma on the server and use the newer version on the client without any fees. However, please keep in mind that this could potentially lead to compatibility issues.
You can use any Nvidia card with architecture equal to or higher than Pascal (in your case it’s Turing, so it fits) on the server; see more on this here.
There is no universal way to calculate the exact number of cameras one GPU can handle, since there are too many variables affecting the outcome (resolution, bitrate, analytics types, etc.). The most straight-forward way to find out is via testing. If you don’t have a T1000 at the moment, it’s possible to test with a different model and than compare the results against any performance table, e.g this one. This will tell you the approximate number of cameras to aim for with your setup.
Q: I’m Looking for a on-prem Camera analytics software for 8 storage bay doors that count and keeps track of Big Bags {2Ton} with different content.
The end user should have accurate insight to the content of the warehouse and the daily content moved in or out.
A: Xeoma Standard and or Xeoma Pro edition are on-premise solutions that can utilize AI video analytics like Object recognition (paid add-on) to detect and count objects through bay doors. Custom development and training of the system’s neural network would be required to specifically identify and track “Big Bags” with different content for detailed reporting. Pricing is here.
To begin with, could you please send us video samples from your cameras and specify the Big Bags there, so that out technical team could assess the task?
By the way, Xeoma can be applied in various scenarios in a warehouse. You can take a look on the in this pdf and on this page. Would you be interested to test Xeoma for any of these tasks? We can provide you with a demo license for any functions.
Q: I installed the latest version xeoma on linux buntu. it has same issue / problem, giving error message: ensuring server part of xeoma in your pc is working.
A: Thank you for providing the error message you faced. You can check solutuions for it here.
In general, if you face any kind of message in Xeoma, you can take a look at possible solutions at Messages in Xeoma page.
October 17, 2025