Performance issues

General discussion about Blue Iris
Cougar281
Posts: 17
Joined: Mon Jan 06, 2020 5:55 pm

Re: Performance issues

Post by Cougar281 »

fuzzkill wrote: Fri Mar 19, 2021 11:11 pm I'm up to 7 5MP cams running 20FPS and all is still well. Have 2 more to put up before I need to order some more.

Might get a 4k to run at the front door, but looks like the system will do everything I need it to do - so I'm happy.

If someone could make an NVR with the configurability and ease of use of BlueIris ... it would kill. Maybe the BI team should work together to port BI to Linux or whatever the NVR's are running for OS.
Well... That has been one of my questions for some time - how can a NVR based on an atom CPU and linux run 32 4MP cameras no sweat but try and run 8 on BI and you need a massive workhorse... To be honest, I think sub stream processing answered that question. I have a strong suspicion that the Dahua NVRs that I'm more or less familiar with, and most likely other NVRS, use sub streams for processing motion data, which is why such low power CPUs are capable of running so many 4MP cameras. I suspect the Dauha NVRs, and most likely other proprietary linux based NVRs operate on a principal similar to the 'new' BI sub stream processing. They use direct to disc recording for the main stream and automatically use sub streams for motion data (where implemented). So my GUESS is you probably wouldn't see huge gains IF BI was ported to linux at this point assuming you're using Direct to disc and sub stream motion processing (although I can't say it would be horrible thing if it WAS ported to linux).
fuzzkill
Posts: 5
Joined: Mon Mar 15, 2021 9:46 pm

Re: Performance issues

Post by fuzzkill »

Cougar281 wrote: Sat Mar 20, 2021 1:31 am
fuzzkill wrote: Fri Mar 19, 2021 11:11 pm I'm up to 7 5MP cams running 20FPS and all is still well. Have 2 more to put up before I need to order some more.

Might get a 4k to run at the front door, but looks like the system will do everything I need it to do - so I'm happy.

If someone could make an NVR with the configurability and ease of use of BlueIris ... it would kill. Maybe the BI team should work together to port BI to Linux or whatever the NVR's are running for OS.
Well... That has been one of my questions for some time - how can a NVR based on an atom CPU and linux run 32 4MP cameras no sweat but try and run 8 on BI and you need a massive workhorse... To be honest, I think sub stream processing answered that question. I have a strong suspicion that the Dahua NVRs that I'm more or less familiar with, and most likely other NVRS, use sub streams for processing motion data, which is why such low power CPUs are capable of running so many 4MP cameras. I suspect the Dauha NVRs, and most likely other proprietary linux based NVRs operate on a principal similar to the 'new' BI sub stream processing. They use direct to disc recording for the main stream and automatically use sub streams for motion data (where implemented). So my GUESS is you probably wouldn't see huge gains IF BI was ported to linux at this point assuming you're using Direct to disc and sub stream motion processing (although I can't say it would be horrible thing if it WAS ported to linux).
Right, the gains aren't for processing. There are crazy cheap NVR's that can do 16 or 32 channels of 4k. Their interfaces just suck. Takes minutes to do what should take seconds. And lack of fine tuning like you can do in BI. And there is no excuse for it. Eventually someone will do it.
ramaz16
Posts: 23
Joined: Sun Mar 14, 2021 4:46 pm

Re: Performance issues

Post by ramaz16 »

fuzzkill wrote: Fri Mar 19, 2021 11:26 am What framerate are you recording/previewing at?
Try detecting motion on substream, and triggering main stream D2D recording ONLY when motion is detected. Play with Trigger options to ensure near 100% motion detection, thus making continuous stream recording for a home system redundant. Also in Cam Settings - Video tick "Limit decoding unless required", and as well try playing with various HW decoding methods.
Post Reply