Nvidia-smi Memory Leak with 570.133.07 and Ubuntu 25.04

Graphics / Linux Linux
1

Since the last Driver/Ubuntu Update, I’m encountering heavy problems with the Graphics Card (4070 Ti SUPER).

  • nvidia-smi allocates all the memory and then crashes without showing any information
  • Suspend does not work properly anymore (UI is fully scrambled when waking up)

The first issue is also discussed here:

https://askubuntu.com/questions/1546016/cant-enable-nvidia-persistence-mode-due-to-nvidia-smi-memory-bug-rtx-2070-kub

and here:

The second here:

The workaround for the suspend issue does not work for me. When I wake up, the screen remains black. I switched to wayland using Ubuntu 25.14. Going back to X11 is also not an option, because WebGL does not work anymore then.

Are these known issues? Thanks.

2

Do you you sudo to start nvidia-smi? I have experienced the issue only when used with sudo.

Regarding suspend, it has never worked with Wayland (on Nvidia cards) AFAIK and now with the latest drivers it appears to be broken even in X11 sessions.

3

Hi @bernhste

I have filed a bug 5301151 for memory leak issue.
Requesting to share nvidia bug report for suspend issue from repro state.

4

Hi @bernhste
Could you please help to check once with beta release: 575.51.02 driver and share test results.

5

Hi @bernhste
Could you please help to check once with beta release: 575.51.02 driver and share test results.
or another possible option would be to rebuild nvidia-persistenced from source:
https://download.nvidia.com/XFree86/nvidia-persistenced/nvidia-persistenced-570.133.07.tar.bz2

6

I encounter the same issue with 570.158.01 and ubuntu 25.04.
Including the bug report.
nvidia-bug-report.log.gz (734.4 KB)

7

instead of using the legacy way of enabling persistence mode nvidia-smi -pm 1 (which is failing) also checked status of nvidia-persistenced systemd service

# systemctl status nvidia-persistenced
● nvidia-persistenced.service - NVIDIA Persistence Daemon
     Loaded: loaded (/usr/lib/systemd/system/nvidia-persistenced.service; static)
     Active: active (running) since Sat 2025-07-19 19:45:33 PDT; 2h 1min ago
 Invocation: 16411c7efda044778730b76ed5e2eac4
    Process: 1905 ExecStart=/usr/bin/nvidia-persistenced --user nvidia-persistenced --no-persistence-mode --verbose (code=exited, status=0/SUCCESS)
   Main PID: 1939 (nvidia-persiste)
      Tasks: 1 (limit: 36516)
     Memory: 360K (peak: 1.6M)
        CPU: 102ms
     CGroup: /system.slice/nvidia-persistenced.service
             └─1939 /usr/bin/nvidia-persistenced --user nvidia-persistenced --no-persistence-mode --verbose

Jul 19 19:45:33 rnv systemd[1]: Starting nvidia-persistenced.service - NVIDIA Persistence Daemon...
Jul 19 19:45:33 rnv nvidia-persistenced[1939]: Verbose syslog connection opened
Jul 19 19:45:33 rnv nvidia-persistenced[1939]: Now running with user ID 120 and group ID 123
Jul 19 19:45:33 rnv nvidia-persistenced[1939]: Started (1939)
Jul 19 19:45:33 rnv nvidia-persistenced[1939]: device 0000:01:00.0 - registered
Jul 19 19:45:33 rnv nvidia-persistenced[1939]: device 0000:05:00.0 - registered
Jul 19 19:45:33 rnv nvidia-persistenced[1939]: Local RPC services initialized
Jul 19 19:45:33 rnv systemd[1]: Started nvidia-persistenced.service - NVIDIA Persistence Daemon.

which is slightly confusing as the daemon is by default configured to run with --no-persistence-mode, so edited that part out and restarted the service which apparently reports no issues and reports “persistence mode enabled”.

# systemctl edit nvidia-persistenced.service --full
Successfully installed edited file '/etc/systemd/system/nvidia-persistenced.service'.
# systemctl daemon-reload
# systemctl restart nvidia-persistenced
# systemctl status nvidia-persistenced
● nvidia-persistenced.service - NVIDIA Persistence Daemon
     Loaded: loaded (/etc/systemd/system/nvidia-persistenced.service; static)
     Active: active (running) since Sat 2025-07-19 21:48:52 PDT; 5s ago
 Invocation: 76aca0176b22464381e8f94685c7d60b
    Process: 11763 ExecStart=/usr/bin/nvidia-persistenced --user nvidia-persistenced --verbose (code=exited, status=0/SUCCESS)
   Main PID: 11765 (nvidia-persiste)
      Tasks: 1 (limit: 36516)
     Memory: 600K (peak: 1.7M)
        CPU: 10ms
     CGroup: /system.slice/nvidia-persistenced.service
             └─11765 /usr/bin/nvidia-persistenced --user nvidia-persistenced --verbose

Jul 19 21:48:52 rnv nvidia-persistenced[11765]: Now running with user ID 120 and group ID 123
Jul 19 21:48:52 rnv nvidia-persistenced[11765]: Started (11765)
Jul 19 21:48:52 rnv nvidia-persistenced[11765]: device 0000:01:00.0 - registered
Jul 19 21:48:52 rnv nvidia-persistenced[11765]: device 0000:01:00.0 - persistence mode enabled.
Jul 19 21:48:52 rnv nvidia-persistenced[11765]: device 0000:01:00.0 - NUMA memory onlined.
Jul 19 21:48:52 rnv nvidia-persistenced[11765]: device 0000:05:00.0 - registered
Jul 19 21:48:52 rnv nvidia-persistenced[11765]: device 0000:05:00.0 - persistence mode enabled.
Jul 19 21:48:52 rnv nvidia-persistenced[11765]: device 0000:05:00.0 - NUMA memory onlined.
Jul 19 21:48:52 rnv nvidia-persistenced[11765]: Local RPC services initialized
Jul 19 21:48:52 rnv systemd[1]: Started nvidia-persistenced.service - NVIDIA Persistence Daemon.

Yet, nvidia-smi still reports the persistence mode to be off.

# nvidia-smi
Sat Jul 19 21:49:09 2025       
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 570.158.01             Driver Version: 570.158.01     CUDA Version: 12.8     |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA GeForce RTX 5070        Off |   00000000:01:00.0 Off |                  N/A |
|  0%   32C    P8              8W /  250W |     235MiB /  12227MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+
|   1  NVIDIA GeForce RTX 4070        Off |   00000000:05:00.0 Off |                  N/A |
|  0%   27C    P8              1W /  200W |      16MiB /  12282MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+
                                                                                         
+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI              PID   Type   Process name                        GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
|    0   N/A  N/A            2411      G   /usr/bin/gnome-shell                    176MiB |
|    0   N/A  N/A            2506      G   /usr/bin/Xwayland                         3MiB |
|    1   N/A  N/A            2411      G   /usr/bin/gnome-shell                      2MiB |
+-----------------------------------------------------------------------------------------+

Unsure, of the discrepancy, and which reporting to trust. nvidia-smi or nvidia-persistenced.

8

Hi All,

Wanted to check if anyone has tried my earlier recommendation, please help to check and share test results.

9

Wanted to check if anyone has tried my earlier recommendation, please help to check and share test results.