Comments (29)
Can you post the log file contents when running with the --debug
option?
from gpu-utils.
I pushed a version that should handle invalid power values read from the gpu, but still need logger details to understand why it is not getting a valid value.
from gpu-utils.
from gpu-utils.
I pushed an update that checks for OSError when reading the pp_od_clk_voltage
file. It was already being checked for all other sensors. I can not verify on my system as an OSError is not raised, even if I read a regular file. To test it out, you will need to clone the repository and run a repository install
Thanks!
from gpu-utils.
from gpu-utils.
I found that I was not not catching OSError for the reading of p-states or the PPM table. I have pushed an update.
from gpu-utils.
from gpu-utils.
Great that you enjoyed some time off during the holiday. I was stuck at home under quarantine since returning from US.
I made some changes to logger during card discovery. It should be more robust now. When you run with --debug
option it should leave a log file in your current directory. Is that not happening in your case?
Also, add added a check of the system type. It should be the first line displayed from gpu-ls
. Later, I will only display in log file and gpu-chk
.
from gpu-utils.
from gpu-utils.
Seems like the log files did not get attached. I am most interested in the log file and output for the latest version that I pushed.
from gpu-utils.
from gpu-utils.
Still not there. I suspect that the Github converter from email to issue update has trouble with complex update. Maybe it will be best to update on the issue page in Github.
from gpu-utils.
The more that I dig into this, the more I realize a rewrite of this section is necessary. I will drop a message here when I complete the work.
from gpu-utils.
I made good progress today. I have added a new command line option to gpu-ls
. The --force_all
option will attempt to read all relevant sensors regardless of the classification of the card or card level readability setting. If a sensor is found to be unreadable, it is added to a disable list. Maybe will tune based on results in your extreme case. It is ready for checkout whenever you have some time.
@csecht Would you like to check it out on your system just to make sure I did not break anything. Also, adding the power dpm state could be useful.
from gpu-utils.
from gpu-utils.
Looks like the code I added to determine system type systemD vs systemV can not run on your system. Seems like os.path.islink throws an error on your system. I will catch it and set system type to unknown. Seems like init
is not a valid command on your system. In debian, it is a link to systemd. It seemed like that was a valid way to determine system type. I will research on a better way.
from gpu-utils.
I completed some major work. Hopefully it addresses all of the issues that we have seen on your distro. I have tested on 4 of my systems with a wide range of GPU models. The --force_all option should now work and will include a list of sensors that it had to disable. I also added a new option for gpu-ls
. The --raw will attempt to read every GPU driver file that it can find and output in a lightly formatted display. This should be useful in exploring read capability of a GPU. The init
exception should now be ok, but still may not be identifying systemV type system. You can test this with gpu-chk
.
from gpu-utils.
from gpu-utils.
I have rewritten significant sections of code to give me more flexibility to manage GPU lists and better manage capability in the various applications. I have tested on my systems with no problems. It should exit more gracefully if you try to run one of the utilities without any GPUs that are compatible.
I only have one APU (first one developed by AMD) so I am curious if your newer generation chip supports more driver capability. Can you run gpu-ls --raw
with the latest in the repository and post the output here? If it does have more capability, then I can update classifications to allow for different levels of APU capability. Thanks!
from gpu-utils.
from gpu-utils.
I have done a major rewrite of the plotting related code. It should now work with all AMD APU and most AMD Legacy GPUs. Can you give it a try? It is the latest in the repository.
from gpu-utils.
from gpu-utils.
Thanks! Seems like all is well. Seems like the attachments by responding by email do not make it into an issue update. It would be cool to see, so post the plot image directly to this issue thread if you get a chance.
from gpu-utils.
from gpu-utils.
I would like to add you to the credits:
@Marjorie-R - Testing, Debug, Verification of AMD-APU Capability
Let me know of any concerns.
from gpu-utils.
from gpu-utils.
from gpu-utils.
On Mon, 2022-05-09 at 06:34 -0700, Rick wrote: Looks like the code I added to determine system type systemD vs systemV can not run on your system. Seems like os.path.islink throws an error on your system. I will catch it and set system type to unknown.
If it's any help runfiles for .deb often include code like this to check for systemd: if [ -d /run/systemd/system ]; then systemctl --system daemon-reload >/dev/null || true if [ -n "$2" ]; then _dh_action=restart else _dh_action=start fi deb-systemd-invoke $_dh_action 'chrony.service' >/dev/null || true fi i.e. in a systemd system then the directory /run/systemd/system should exist. On my PC with Devuan, where I use sysvinit instead, it is absent.
…
-- Marjorie
I have implemented a check that I derived from stack overflow. Can you try ./gpu-ls --verbose
. System type should be one of the first items displayed. Not sure if I asked you to check this in the past.
from gpu-utils.
This should be resolved.
from gpu-utils.
Related Issues (20)
- Feature Request: Support values from sysfs gpu_metrics
- index out of bound error HOT 4
- Can not read parameter and list index out of range HOT 5
- Tool doesn't see GPUs when PCI address has multiple domains HOT 9
- Add GDDR6X memory temps to the sensor readouts for Nvidia cards
- Ubuntu 22.04 Debian Package - Invalid Icon Path HOT 1
- Warning: apt-key is deprecated in Ubuntu 22.04 HOT 1
- env.py: conversion of hexadecimal to int fails HOT 2
- Wrong vendor for on-board GPU HOT 1
- AMD Ryzen 5 5700G with Cezanne not working HOT 5
- gpu-mon requires gtk modules even when --gui is not used HOT 10
- ValueError: invalid literal for int() with base 10 at GPUmodules.py line 1236 in read_gpu_pstates HOT 10
- RX7900 (gfx11) Cards Fan Control is Not Functional HOT 10
- When using gpu-pac I receive an error HOT 2
- Support for MacOS-X ? HOT 2
- ValueError: invalid literal for int() with base 10: HOT 12
- VRAM and GTT memory usage reported as None HOT 5
- Q: had it been tested w/ latest Debian bpo Kernel? HOT 14
- PermissionError: [Errno 1] Operation not permitted HOT 9
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from gpu-utils.