After problems with Debian package, I tried to build it from the source by following the article from here https://www.linuxuprising.com/2019/06/2-tools-for-monitoring-nvidia-gpus-on.html . I was able to build qnvsm but while running the command
"Connecting to nvidia-smi...
Segmentation fault (core dumped)"
error is coming

There are missing C+G processes in the list. Regexp in

I have changed the regexp in 14822ca and it looks ok now:

Hi,

Sorry for bothering you, but I'm having issues getting accurate utilization from NVML's process structs for my own Nvidia GPU monitoring utility. It seems like your tool is far more accurate than mine(e.g. mine reports 0%, yours reports 12%) but I'm not sure why that is and was hoping you'd be willing to share how you managed to get such accurate utilization numbers.

Mine updates about every 500ms in the backend but frontend only updates about 1000ms and I'm using the timestamp from the previous update period as the function timestamp input.

Using system:

Linux Mint 19.1
Linux kernel 4.15.0-136-generic
cmake 3.10.2
qtdeclarative5-dev 5.9.5-0ubuntu1.1
nividia-smi 455.45.01

Steps to repro:

1. git clone https://github.com/congard/nvidia-system-monitor-qt
2. cd nvidia-system-monitor-qt
3. mkdir build
4. cmake -DCMAKE_BUILD_TYPE=Release -DIconPath=/usr/share/icons/hicolor/512x512/apps/nvidia-system-monitor-qt.png -B build -G "Unix Makefiles"

Error output:

CMake Error: The source directory "/mnt/Data/Repositories/nvidia-system-monitor-qt/build" does not appear to contain CMakeLists.txt.
Specify --help for usage, or press the help button on the CMake GUI.

Description:

CMakeLists.txt obviously exists, but cmake is apparently looking for it in the wrong place and I can't figure out why. I tried adding an "-S ." parameter to the command to specify the local folder (repo root) as the source root, but no change in error output. You can copy CMakeLists.txt to the build folder obviously, at which point the command finds it, but then the paths are messed up and it can't find the source files and configure breaks. Alternatively you can change the -B parameter to just "." and configure completes successfully, but then the outputs dump to the repo root rather than the build folder and it's unclear what additional steps need to be taken to be able to continue with the last installation step.

My question is not related to wifimouse project.

I download the app wifimouse for windows.

wifimouse app fully build with qt.

Source code of wifimouse server application

I install and run wifimouse app in my computer it works perfectly and all the options are works smoothly.

But, If I run the same code in my computer it not works properly.

For example ; In wifimouse application there is an option for brightness controling

Brightness code

error comes like this. But , it works smoothly in downloaded installer from here

Problem with qt in getting outputs.

My project

I want to get the cpu usage , ram usage ,process list and app list with qt.

In Windows

I am getting ram usage from this code.

But app list , cpu usage and process outputs are not showed.

In linux

In linux the cpu usage output not showed.

In Mac

In mac the app list output not showed.

Why this problem occured ?

Resizing the window doesn't resize the graph (at least vertically) which would be very handy.

Hi, this command gives me gpucount=8 while starting build/qnvsm gives

the monitor is not showing anything on the Utilization tab (the process tab seems working)


Any idea about this please could you let me know?
Thank you.

I want to compile my code with Compile: qmake && make.

https://github.com/PradeepSimba/test/blob/main/.github/workflows/main.yml

But , If I run it comes an error .

https://github.com/PradeepSimba/test/runs/6637488143?check_suite_focus=true#step:7:59

How can I solve this error ?

I'm on linux 18.04 LTS

(base) pomp@pomp:~$ cd nvidia-system-monitor-qt/
(base) pomp@pomp:~/nvidia-system-monitor-qt$ qmake qnvsm.pro
(base) pomp@pomp:~/nvidia-system-monitor-qt$ make
g++ -c -pipe -O2 -Wall -W -D_REENTRANT -fPIC -DQT_NO_DEBUG -DQT_WIDGETS_LIB -DQT_GUI_LIB -DQT_CORE_LIB -I. -I../anaconda3/include/qt -I../anaconda3/include/qt/QtWidgets -I../anaconda3/include/qt/QtGui -I../anaconda3/include/qt/QtCore -Ibuild -I../anaconda3/mkspecs/linux-g++ -o build/utilization.o src/utilization.cpp
src/utilization.cpp: In function ‘void drawGrid(QWidget*, QPainter*)’:
src/utilization.cpp:33:13: error: ‘class QFontMetrics’ has no member named ‘horizontalAdvance’
     x0 = fm.horizontalAdvance(text);
             ^~~~~~~~~~~~~~~~~
src/utilization.cpp:37:13: error: ‘class QFontMetrics’ has no member named ‘horizontalAdvance’
     x0 = fm.horizontalAdvance(text);
             ^~~~~~~~~~~~~~~~~
src/utilization.cpp: In function ‘void drawStatusObjects(std::vector<QRect, std::allocator<QRect> >&, UtilizationData*, QPainter*)’:
src/utilization.cpp:74:24: error: ‘class QFontMetrics’ has no member named ‘horizontalAdvance’
     int textWidth = fm.horizontalAdvance("100%"); // because max width will be only at 100%
                        ^~~~~~~~~~~~~~~~~
Makefile:1726: recipe for target 'build/utilization.o' failed
make: *** [build/utilization.o] Error 1

Hi,

I installed the flatpak release on Ubuntu Mate 22.04 and I don't see any charts in the GUI.

I first thought that maybe it's the pink line since the GPUs sit at 0% utilization, but even when I maxxed all of them to 100% the line didn't move. Any idea what could be causing this?

UPDATE
Mystery solved - there are so many GPUs that the charts got squished into what looks like a line. When I vertically expanded the window, the charts appeared as expected. I don't know how difficult this would be in QT, but would it be possible to make the window scrollable? Right now I cannot access the bottom values since I have to make the window taller than my screen to see the charts at all, but that pushes the stats outside the screen area.

Hi,

In Win10 resource manager you can see GPU decoder/encoder % usage, among other things. Would it be possible to track that as well?

The memory usage tab displays nvidia-smi's utilization.memory value, which is "
Percent of time over the past sample period during which global (device) memory was being read or written" (from the useful queries page) and not the amount of vRAM in use.

I changed

to

        memoryData[i - 1].total = std::atoi(split(data[1], " ")[0].c_str());
        memoryData[i - 1].free = std::atoi(split(data[2], " ")[0].c_str());
        memoryData[i - 1].used = std::atoi(split(data[3], " ")[0].c_str());
        udata[i - 1].level = memoryData[i - 1].used * 100 / memoryData[i - 1].total;

in my local build to make the tab display percent of vRAM in use

Following the instructions to compile from scratch I get an error when I run make:

Scanning dependencies of target qnvsm
[ 33%] Building CXX object CMakeFiles/qnvsm.dir/qnvsm_autogen/mocs_compilation.cpp.o
[ 33%] Building CXX object CMakeFiles/qnvsm.dir/src/mainwindow.cpp.o
[ 44%] Building CXX object CMakeFiles/qnvsm.dir/src/main.cpp.o
[ 55%] Building CXX object CMakeFiles/qnvsm.dir/src/processes.cpp.o
[ 66%] Building CXX object CMakeFiles/qnvsm.dir/src/utilization.cpp.o
[ 77%] Building CXX object CMakeFiles/qnvsm.dir/src/utils.cpp.o
[ 88%] Building CXX object CMakeFiles/qnvsm.dir/src/worker.cpp.o
/home/Documents/nvidia/nvidia-system-monitor-qt/src/utilization.cpp: In function ‘void drawGrid(QWidget*, QPainter*)’:
/home/Documents/nvidia/nvidia-system-monitor-qt/src/utilization.cpp:32:13: error: ‘class QFontMetrics’ has no member named ‘horizontalAdvance’
x0 = fm.horizontalAdvance(text);
^~~~~~~~~~~~~~~~~
/home/Documents/nvidia/nvidia-system-monitor-qt/src/utilization.cpp:36:13: error: ‘class QFontMetrics’ has no member named ‘horizontalAdvance’
x0 = fm.horizontalAdvance(text);
^~~~~~~~~~~~~~~~~
/home/Documents/nvidia/nvidia-system-monitor-qt/src/utilization.cpp: In function ‘void drawStatusObjects(std::vector<QRect, std::allocator >&, UtilizationData*, QPainter*)’:
/home/Documents/nvidia/nvidia-system-monitor-qt/src/utilization.cpp:73:24: error: ‘class QFontMetrics’ has no member named ‘horizontalAdvance’
int textWidth = fm.horizontalAdvance("100%"); // because max width will be only at 100%
^~~~~~~~~~~~~~~~~
CMakeFiles/qnvsm.dir/build.make:114: recipe for target 'CMakeFiles/qnvsm.dir/src/utilization.cpp.o' failed
make[2]: *** [CMakeFiles/qnvsm.dir/src/utilization.cpp.o] Error 1
CMakeFiles/Makefile2:76: recipe for target 'CMakeFiles/qnvsm.dir/all' failed
make[1]: *** [CMakeFiles/qnvsm.dir/all] Error 2
Makefile:83: recipe for target 'all' failed
make: *** [all] Error 2

I installed sudo apt install qtdeclarative5-dev but it seems QT doesn't like the horizontalAdvance member...

Any clues?

Thanks

Can you please add the following UI element options IF possible.

• dark mode
• nonstacked graphs
• smooth graphs

Dark mode would be the most important as it would please the eyes.

Attempting to launch it gives this error:

qnvsm: /lib/x86_64-linux-gnu/libQt5Core.so.5: version `Qt_5.15' not found (required by qnvsm)

I opened nvidia-system-monitor-qt and nvidia-smi side-by-side to compare values, and for some reason the GPU memory use values per process are significantly smaller in this application than in the nvidia-smi cli interface. I have attached a screenshot below.

Which one should I trust? Is this something to do with the way this application retrieves the values from nvidia-smi?

ZeevoX

Nvidia smi usually displays this information in its output by default. Would it be useful to include this information as well ?

If so, willing to work on adding it but would like to know if you have an opinion on which tab it should go on and should it be displayed as a graph or point in time ?

Any chance of publishing the rpm builds to https://copr.fedorainfracloud.org/ or similar?