No memory on server side
I’m trying to use spacedesk on my PC, and when I try to enable it I get “No memory on server side” message.
I’m currently using 4 screens directly on my GPU, and the tablet with spacedesk would be the 5th.
I have a Ryzen 9950X3D and a RTX5090 with 64GB RAM. It’s currently only using 10% RAM and 10% VRAM, so the “no memory” message is weird.
Uploaded files:- You need to login to have access to uploads.
Hi @rafael-3,
For further analysis, please send us the logs of your Primary PC by opening the spacedesk Driver Console as administrator and clicking on Diagnostics then follow the steps below:
1. Download the DebugView.
2. Check the Diagnostic Collection box.
3. Reproduce the issue e.g. by connecting the spacedesk Viewer app.
4. Uncheck the Diagnostic Collection box.
5. Click “Save All Information” button (this might take a while to finish).
6. Compress/Zip the folder which contains the logs, then attach to your next reply to this post. If the folder exceeds 512kb, you may upload it to Dropbox or Google Drive then include the download link on your next reply.
Thank you.
Similar issue here, using Ryzen 9950X and RTX5080. VRAM leak on server GPU.
I would like to add:
1. Issue appears inconsistent, it can occur 10mins into a session, or 4 hours into a session, making it difficult to reproduce reliably.
2. The VRAM leak is not reflected properly on task manager and other monitoring software. They all show very low/normal VRAM usage. The VRAM leak is only noticed in nvidia-smi. Even with continuous logging flag(-l 1), it still sometimes does not track the used VRAM properly (but the impact of VRAM leak is observed such as in (3)).
3. VRAM leak causing system to fallback to OpenGL 1.1, GUI related applications crash or unable to start (file explorer, snipping tool, gpu-z, gpu-shark)
4. VRAM may or may not be released when session ends, and may or may not be released even after terminating spacedesk processes on server. On multiple testing, VRAM is sometimes released and other times not.
5. I’ve attempted DDU-ing and trying different version of NVIDIA drivers (Game ready 595.71, 591.74, 591.59, and studio 591.74), attempted to reinstall spacedesk driver as well, but the issue still happen since January’s update.
As mentioned in (1), it is quite difficult and time consuming to replicate and troubleshoot due the unpredictable time before leak starts. I will update again when I have more logs or information.
@spacedesk_nicole, do you just need the .log files, or .txt files too? The entire folder contains multiple dumps and etl files, making the compressed folder still very large (300mb+).
I’ve included a log of gpu vram in the folder. This was the smallest (shortest time) I could replicate this issue.
https://drive.google.com/drive/folders/1RvnjKsJCVYPN1J45ntq9BTilCS1teNIP?usp=sharing