Best Practices for Optimizing ZoneMinder 1.36.x Performance with Multiple High-Resolution Cameras

[Quantum]

{... and everyone holds their breath ...}

At least turn off jpg. And set one monitor for the hi stream and one for lo. Set hi to NoDect and lo to MoCord. Link hi to lo so it triggers when lo does; hi does not analyze, it only records when lo triggers. Set hi to Camera Passthrough and lo to Encode.

This saves enormous CPU, not analyzing nor even decoding the hi stream, but unfortunately for events does not record object bounding boxes (if you can ever get ML working).

Turn on date/time/text in the cameras, not in ZM. Be sure to get your resolutions right. Check system status using nmon to see whether you're CPU/RAM/Disk-bound.

This is as far as I've gotten.

[Andyrh]

Camera FPS is important. You are not making a movie so set the FPS to around 5-10 on the cameras. Also, I set the keyframe to 1 per second.
Not a hard rule, but 1 thread per motion detection works best in my experience.
Camera passthrough is important. This reduces CPU work.
CPU clock rate is important after you have enough threads. Or said another way, I prefer cores over clock.
I have seen many threads about RAM. I have 6 modect cameras and 6 recording with 16GB of RAM. I frequently have 8GB or more free for use.
Hardware RAID controller helps, but may be overkill on my part.
Good HDDs, I like the WD Purples. Any good HDD will do, just avoid the green ones, they cannot take the abuse.

[mikb]

Original posting was an AI generated/stolen content spammer posting links to gaming sites ...