Page 1 of 1
requirements for up to 10 cameras
Posted: Thu Jun 23, 2011 3:16 pm
by developer123
Hey
I'm planning to install a system with up to 10 network cameras in 1024x768 resolution with somewhere around 24fps.
Which hardware-specs am I looking at if I want to use motion-detection with zones?
Does ZoneMinder use the graphics card, or do I just need a good CPU?
Would it help to use a quad-core?
Does it make sense to run the system on 64bit?
How much RAM would I need?
many question, but I hope you can help me. I wouldn't mind if you suggest some mainboards/CPUs as well
thanks
Re: requirements for up to 10 cameras
Posted: Fri Jun 24, 2011 5:28 am
by Flasheart
Yes, go quad. ZM benefits a lot from multiple cores. An HP ML110 or ML115 quad would do this fine, with one exception - and aim for 4gb of ram. No massive need for 64 bit. Standard 32b kernels won't be able to access > 3.5 gb of ram unless you choose a bigmem kernel, but that's easy.
The exception btw is 24fps. That's ridiculous and almost certainly a) Not achievable in a sustained manner from megapixel ip cams and b) not required. Under 5fps is the norm for recording and motion detection.
Re: requirements for up to 10 cameras
Posted: Sat Jun 25, 2011 3:21 am
by theforce
I'm running 10 IP cameras at 640x480 and 5fps with motion detection. I'm running this on an AMD 4 core 3ghz, 2GB ram, and Ubuntu 64bit. The average load on the system is about 2.5. I figure I can do another 4-6 cameras on this setup just fine with the load staying under 4. I bought the mobo/cpu/mem on newegg.com for under $180.
Re: requirements for up to 10 cameras
Posted: Sat Jun 25, 2011 1:44 pm
by developer123
@Flasheart I understand, the framerate will be trouble for motion detection. The problem is I want to play the video itself smoothly afterwards. Could it be possible to record the video with a high framerate, but let ZoneMinder only compare frames (detect motion) with 5fps. So something like "compare every 5th frame"? So in the end I'll get a good video, but still profit from the less calculation?
Converting all videos down to 5fps, run motion detection and then cut the snippets out of the original video, is a waste of resources isn't it?