My instance is running on a virtual machine. The issue is present even after a fresh reinstall on a new VM.
This only appears to be an issue with newer cards. The Jan release works fine with a Quadro k1200/4gb (BlueIris/W10 system) but when swapped for a T600/4gb (Turing based card similar to a GTX1650), it stops detecting objects. Everything is the same (driver, cuda version 11.5 and cudnn 8.3.3). In that case, uninstalling DS and going back to the Sept 2021 release fixes the issue.
I can report something similar as vulture1. Using linux OS with a P400 and it works just fine. When I switch the card to a T600, it stops working. Nothing changes in between.
I have the same problem. I changed my old GTX 650 which didn’t work with Deepstask-GPU to Nvidia T600, it works, but doesn’t return any objects to “Agent DVR”.
Same problem here… worked fine with CPU version, then installed a Nvidia GFX 1660S and upgraded to the gpu docker image and now it returns 200 but no object detected.
Downgraded to gpu-2021.09.1 and now it works
@hflatoey Thanks for reporting this. We noticed latest DeepStack GPU works on some GPUs and doesn’t on others. We traced this to recent bump in the PyTorch version which caused a number of issues on different CUDA versions… As mentioned in an earlier post in this thread, we are working on multiple DeepStack GPU version releases that will target different CUDA versions, ensuring compatibility with your individual graphics cards.
Hello, do you have a fix for the GPU version to work on Blue Iris? I have a Quadro T400 GPU and in version 2022.01.01 it doesn’t work at all. Only in version 2021.09.01 it works very well.
I would like to use a newer, better-identified version.
Yes, we’ve all been waiting for months. You can see my earlier posts on this. I believe that if you use an older version of CUDA the GPU version may work with the current DeepStack. You can read the DeepStack installation instructions to get that version number. It might or might not be available due to the more recent CUDA releases.
The last I heard the authors were working on a fix for using current/multiple CUDA versions. However, that has been months and my guess (not having looked at the code) that that wouldn’t take all that long.
Thank you for this!! I’ve been banging my head against the wall trying to get this to work for weeks. I finally have deepstack working on my GPU by rolling back to the 2021.09.1 and I finally have deepstack running on my GPU. I see they have a fix coming so I’ll keep an eye out and wait for that.
Good morning, any resolution or version that works on GPU? I’m looking forward to updates.
I am getting no results from GPU Docker on Ubuntu LTS 22.
NVIDIA-SMI 515.65.01 Driver Version: 515.65.01 CUDA Version: 11.7
I see a defunct REDIS server. I get a response 200, but no predictions.
tried deepstack:gpu and :gpu-2022.01.1 no difference.