AMD vs Intel for BI?

CCTVCam

Known around here
Joined
Sep 25, 2017
Messages
2,674
Reaction score
3,505
Not that far off building hopefully. Just wondering what the above situation was these days AMD vs Intel?

I know the Wiki still says Intel, however, I seem to remember several threads that say these days don't use HW acceleration in BI, in which case if you're not using Quicksync to do the encoding, but integrated graphics or an internal card,does it matter anymore if it's Intel or Amd?
 
Last edited:

Shap

Young grasshopper
Joined
Jul 4, 2021
Messages
37
Reaction score
22
Location
Austin,TX
I think you do use HW acceleration in BI, nowadays especially. So I guess the answer is you better with Intel with integrated graphics for hardware acceleration. While you can use GPU for that it is less efficient by far (according to the same wiki at least)
 

sebastiantombs

Known around here
Joined
Dec 28, 2019
Messages
11,511
Reaction score
27,695
Location
New Jersey
If you really want to minimize CPU utilization hardware acceleration is a must. Granted it does not reduce CPU utilization as much as sub streams, but it still reduces it.

A secondary GPU, NVidia specifically, can be useful if you use DeepStack integration. It will allow much faster and more accurate analysis by DeepStack. The power penalty is not particularly high in this case, under 20 watts. Using an NVidia card for hardware acceleration will consume more power, on the order of 100 watts or so.

Best combination if you're running BI and DeepStack is an Intel processor with QuickSync and an NVidia card that is CUDA capable like the Quadro series.
 

CCTVCam

Known around here
Joined
Sep 25, 2017
Messages
2,674
Reaction score
3,505
Thanks. So presumably if you want AI, you're better off buying a camera with AI built in than using Deepstack as a 3rd party add on.
 

sebastiantombs

Known around here
Joined
Dec 28, 2019
Messages
11,511
Reaction score
27,695
Location
New Jersey
Camera AI, decent camera AI, gets expensive, figure another hundred bucks, or more, per camera. DeepStack in BI is free and can be used on any "dumb" camera. Camera AI, to work properly also needs an NVR with AI capability. An NVR is not as flexible as BI and is relatively limited in capability when compared to BI on a PC. It's a trade off of what you want, are willing to pay for and the overall flexibility that you may need.
 

Rob2020

Getting comfortable
Joined
Oct 2, 2020
Messages
996
Reaction score
2,574
Location
OR USA
I built a system around a Ryzen 2700 & 16GB RAM far before I decided to get involved in IP Cams, it runs solid and stable with low CPU usage.

With 4 Dahua cams (Mix of 2MP/4MP) recording 24/7, CPU usage stays below 10%.

I have an Nvidia Super GT1650 card which shows zero CPU in task manager so it appears the Ryzen 2700 is doing all the work.

I have not played around with trying to put the GPU into the mix.

Intel seems to be the answer based upon all my reading but AMD will definitely work.
 
Top