MikeLud1
IPCT Contributor
To confirm if it is a BI or CP.AI issues open up CP.AI Explorer and run a test, if it works using CP.AI Explorer then its a BI issues.
Explorer Link
CodeProject.AI Version 2.0
- Thread starter MikeLud1
- Start date
Explorer Link
Vettester
Getting comfortable
- Joined
- Feb 5, 2017
- Messages
- 748
- Reaction score
- 704
Interesting... I'm running the latest version of BI (5.7.6.2) with CPAI 2.1.8 and I'm not seeing any issues. When I updated CPAI I did not uninstall the previous version (2.1.6).
Last edited:
swilliams76360
Getting the hang of it
Im running the latest BI and CPAI 2.1.8 clean install of CPAI reinstalled yolov5 6.2 and facial rec and absolutely zero issues the new version of both seems to be faster and is recognizing the faces alot quicker and getting more positive results on faces...good job all...
cyberwolf_uk
Getting comfortable
- Joined
- Sep 27, 2014
- Messages
- 611
- Reaction score
- 725
Just for info, I upgraded from CP 2.1.6 to 2.1.8 without any uninstall and everything worked as expected. All I did was close down BI before the update to 2.1.8. No reboot at any time so like @Vettester that two of us it worked for.
misterfredsr
Getting the hang of it
I may try updating this weekend, I'm still on 1.6.7.
lbphotoguy
n3wb
Anyone know if CPAI 2.1.8 solves the issue of recognizing NVIDIA CUDA graphics cards? 2.1.6 had issues for some of us with NVIDIA cards where it would only run in CPU mode, not GPU (CUDA) mode with YOLOv5 6.2 . By downgrading to 2.0.8 was the temporary fix to running in GPU (CUDA) mode.
cyberwolf_uk
Getting comfortable
- Joined
- Sep 27, 2014
- Messages
- 611
- Reaction score
- 725
I have a gt1030 running in my box and it works fine with GPU running YOLOv5.6.2 on 2.0.8, 2.1.6 & now 2.1.8. I must admit it did have all sorts of issues with detection etc on my test box running the same setup until I did a clean install of everything, I've mentioned it in a thread somewhere on here.
Here are the versions
cyberwolf_uk
Getting comfortable
- Joined
- Sep 27, 2014
- Messages
- 611
- Reaction score
- 725
For me .net process at roughly 250ms and 5.6.2 between 60 - 120ms with my GT1030
Hmmm, when I switched from 6.2 to net I got under 100ms on my 970 GPU. Course I am still running 2.0.8-Beta. When I upgrade I will have to play with them, thanks.
cyberwolf_uk
Getting comfortable
- Joined
- Sep 27, 2014
- Messages
- 611
- Reaction score
- 725
Strange thing is when I enable hardware acceleration it jumps up to the same as .net.... confused.com 
I have not even changed anything there, that I remember, so these are the defaults:Strange thing is when I enable hardware acceleration it jumps up to the same as .net.... confused.com
View attachment 161954
lbphotoguy
n3wb
I have NVIDIA GTX 1660 Super running CUDA 12.1 on 2.0.8 properly. Will try 2.1.8 this weekend. Thanks for your feedback.
misterfredsr
Getting the hang of it
I have the 1660 also and I'll be doing the update on Sunday. Saturday spending the day with granddaughter planting flowers.
cyberwolf_uk
Getting comfortable
- Joined
- Sep 27, 2014
- Messages
- 611
- Reaction score
- 725
Make sure you run the cuDNN install script, I had to run the script again when I upgraded to 2.1.6 despite my GPU working in previous versions.
File Download - CodeProject
www.codeproject.com
lbphotoguy
n3wb
Good tip, was not planning on running cuDNN again. Will do. Will also uninstall CPAI 2.0.8, delete the directory, reboot, then install CPAI 2.1.8.Make sure you run the cuDNN install script, I had to run the script again when I upgraded to 2.1.6 despite my GPU working in previous versions.
File Download - CodeProject
So far it does seem a little faster:
Prior to change:
My CPU % shot up:
But in BI all looks good, will need to monitor it some:
Nevermind, look at the above after change, looks like it went Offline