-
-
Notifications
You must be signed in to change notification settings - Fork 250
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Wrong display when GPU memory is just a little bit over 1G #413
Comments
just noticed that page 4MEM shows the same as well:
|
Hi @clogwog (also reply for @artur-ag and @vesselofgod #469 ) I am aware of an issue where the output plotting in my jtop new version generates errors. I have worked to reduce the occurrence of this bug. The issue stems from the use of ASCII characters to plot the output, and there are cases where there is no small or big ASCII block available to perfectly plot the output. I'm not sure if you have any experience with ASCII coding, but I was wondering if you could help me fix it. It would be really helpful for me! I wrote and updated this code some time ago, and the plot object is available at https://github.com/rbonghi/jetson_stats/blob/master/jtop/gui/lib/chart.py |
From what I understand, the issue is that there's only space for 4 characters in most places of the UI, but all values between One way of fixing it would be to start using the higher unit whenever the value goes above 999, even if the value in the higher unit is not above 1 of the larger unit. Like this:
@rbonghi I'm not familiar with the code, but from what I can tell, it's not chart.py making these conversions between units, but some other code before this, right? I see common.py has some unit conversion code. Maybe that's where this new behavior needs to be implemented (if you agree that this solution is adequate, that is). |
Describe the bug
When GPU memory goes up to 1 G it shows it wrongly as 101M for a bit.. Once it gets higher i think I have seen it display 1G properly.. but it must be around the change-over from M to G.
This confused me a bit because i thought i was running light on memory, until i saw the startup sequence as shown in the video below and confirmed it with the graph on page 2(GPU)
To Reproduce
When starting a program that will take heaps of GPU memory it goes up to 900M then it suddenly shows 101M even though the memory has still gone up.
It shows it wrong in both the 1(All) screen next to the process as well in the 2(GPU) total GPU graph
See this example https://www.youtube.com/watch?v=R_rWsqLXMfw
Expected behavior
just showing either 1000 M or 1 G
Board
Output from
jetson_release -v
:You can find this data on:
jetson_release -v
jtop (page INFO)
-->
jetson-stats version: [e.g. 1.8]
P-Number: [e.g. pXXXX-XXXX]
Module: [e.g. NVIDIA Jetson XXX]
Jetpack: [e.g. 4.3]
L4T: [e.g. 5.2.1]
Log from jtop.service
Attach here the output from:
journalctl -u jtop.service -n 100 --no-pager
Log from jetson-stats installation
Attach here the output from:
sudo -H pip3 install --no-cache-dir -U jetson-stats
The text was updated successfully, but these errors were encountered: