| Summary: | Total GPU VRAM sensor reports value for NVIDIA GPU that differs from what nvidia-smi reports | ||
|---|---|---|---|
| Product: | [Frameworks and Libraries] ksystemstats | Reporter: | euler <asudox+7frg23vcr> |
| Component: | General | Assignee: | Plasma Bugs List <plasma-bugs-null> |
| Status: | RESOLVED DUPLICATE | ||
| Severity: | normal | CC: | ahiemstra, john.kizer, kde, nate, plasma-bugs-null |
| Priority: | NOR | ||
| Version First Reported In: | 6.3.2 | ||
| Target Milestone: | --- | ||
| Platform: | Arch Linux | ||
| OS: | Linux | ||
| Latest Commit: | Version Fixed/Implemented In: | ||
| Sentry Crash Report: | |||
| Attachments: |
nvidia-smi --query results
Similar wrong reporting, looks like FB + BAR1 too |
||
|
Description
euler
2025-03-02 11:20:48 UTC
Hi - I can reproduce this on my Fedora KDE 41 device with Plasma 6.3.3. nvidia-smi reports ~12.3GiB, while System Monitor reports 28.0GiB. Thanks! Please attach output of nvidia-smi --query Created attachment 180323 [details] nvidia-smi --query results Nice call, I think that's it - it looks like the FB and BAR1 Memory Usage numbers are being summed together, although from my very limited understanding (https://man.archlinux.org/man/nvidia-smi.1.en), they are two different ways of mapping the same actual memory? Sorry for the multiple emails, mis-click on my end on the status! Created attachment 186316 [details]
Similar wrong reporting, looks like FB + BAR1 too
I'm observing similar behavior so I'm posting my values here to help with diagnosis.
Reported RAM looks like it is adding BAR1 values (as seen on screenshot).
GPU: NVIDIA GeForce RTX 3080 Lite Hash Rate
DE: KDE Plasma 6.4.5
WM: KWin (Wayland)
Kernel: Linux 6.17.1-2-cachyos
*** This bug has been marked as a duplicate of bug 510518 *** |