gpustat | simple command-line utility | Monitoring library

by wookayin Python Version: 1.1.1 License: MIT

X-Ray Key Features Code Snippets(5)Community Discussions(5)Vulnerabilities Install Support

kandi X-RAY | gpustat Summary

gpustat is a Python library typically used in Performance Management, Monitoring applications. gpustat has no bugs, it has no vulnerabilities, it has build file available, it has a Permissive License and it has medium support. You can install using 'pip install gpustat' or download it from GitHub, PyPI.

[license] Just less than nvidia-smi?. Self-Promotion: A web interface of gpustat is available (in alpha)! Check out [gpustat-web][gpustat-web].

Support

Quality

Security

License

Reuse

Support

gpustat has a medium active ecosystem.

It has 3417 star(s) with 261 fork(s). There are 44 watchers for this library.

There were 1 major release(s) in the last 12 months.

There are 21 open issues and 86 have been closed. On average issues are closed in 304 days. There are 5 open pull requests and 0 closed requests.

It has a neutral sentiment in the developer community.

The latest version of gpustat is 1.1.1

Quality

gpustat has 0 bugs and 20 code smells.

Security

gpustat has no vulnerabilities reported, and its dependent libraries have no vulnerabilities reported.

gpustat code analysis shows 0 unresolved vulnerabilities.

There are 0 security hotspots that need review.

License

gpustat is licensed under the MIT License. This license is Permissive.

Permissive licenses have the least restrictions, and you can use them in most projects.

Reuse

gpustat releases are available to install and integrate.

Deployable package is available in PyPI.

Build file is available. You can build the component from source.

Installation instructions, examples and code snippets are available.

gpustat saves you 316 person hours of effort in developing the same functionality from scratch.

It has 760 lines of code, 57 functions and 5 files.

It has high code complexity. Code complexity directly impacts maintainability of the code.

Top functions reviewed by kandi - BETA

kandi has reviewed gpustat and discovered the below as its top functions. This is intended to give you an instant insight into gpustat implemented functionality, and help decide if they suit your requirements.

Prints the GPU stats
Print the output to fp
Print the output
Write the object to fp
Return a JSON representation of the group
Return a list of keys
Creates a new query
Write the summary of the report
Write msg to stderr
Adds an error to the report
Removes all processes from GPU
Run a new query
Runs the Twine package
Print a status message
Loop over the screen
Print the GPU stats
Return the version string
Read the contents of the README md file

Get all kandi verified functions for this library.

gpustat Key Features

No Key Features are available at this moment for gpustat.

gpustat Examples and Code Snippets

guppy_parameter_optimiser,quick look at results,FAST model

Shell

Lines of Code : 21

License : Strong Copyleft (GPL-3.0)

Copy

grep -o 'chunks per runner: .*\|samples/s:.*' param_sweep_test/guppy_fast_*

param_sweep_test/guppy_fast_160.out:chunks per runner:  160
param_sweep_test/guppy_fast_160.out:samples/s: 3.30911e+07
param_sweep_test/guppy_fast_256.out:chunks per runner:

Install

Shell

Lines of Code : 5

License : No License

Copy

wget https://raw.githubusercontent.com/Fangyh09/gpustatus/master/gpustatus.sh
chmod 755 gpustatus.sh

# optional 
sudo ln -s gpustatus.sh /usr/local/bin/gpustatus

Usage

Shell

Lines of Code : 3

License : No License

Copy

1. bash gpustatus.sh or 
2. ./gpustatus.sh or
3. gpustatus (if have linked it to /usr/local/bin/gpustatus)

How to utilize 100% of GPU memory with Tensorflow?

Python

Lines of Code : 4

License : Strong Copyleft (CC BY-SA 4.0)

Copy

pip install gpustat

gpustat -i

when training simple code of pytorch, cpu ratio increased. GPU is 0% approximately

Python

Lines of Code : 10

License : Strong Copyleft (CC BY-SA 4.0)

Copy

# Configure your device
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
# Upload your model onto GPU
net.to(device)
# Upload your tensor onto GPU
inputs, labels = inputs.to(device), labels.to(device)

Community Discussions

Trending Discussions on gpustat

OOM when allocating tensor with shape[3075200,512] and type float on : why?

Im not running anything on GPU. But GPU stat is not clearing my usage

Why can't I run tensorflow session on CPU while one GPU device's memory is all allocated?

Is it possible to train a H2O model with GPU and predict with a CPU?

OpenGL program with tensorflow C++ gives failed call to cuInit : CUDA_ERROR_OUT_OF_MEMORY

QUESTION

OOM when allocating tensor with shape[3075200,512] and type float on : why?

Asked 2020-Oct-04 at 05:42

I am experimenting by creating a grayscale image to use the CNN model(using keras). I have to experiment with the image size of 512*512, but I get a memory overflow error. I can't reduce the image size anymore, and if I look at gpustat, it's 32GB, but I don't know why the memory error pops up. I am not sure where and how to find and increase the allocable gpu memory.

Here are the attempts I have made so far.

1.Reducing the number of neurons in a layer 2. Using a smaller batch size (current batch size 5) 3. Increase Maxpooling size to use 4. Increase stride size 5. Dropout 6. config = tf.ConfigProto() config.gpu_options.allow_growth = True 7. config = tf.ConfigProto() config.gpu_options.per_process_gpu_memory_fraction 8. strategy = tf.distribute.MirroredStrategy() 9. mirrored_strategy = tf.distribute.MirroredStrategy(devices=["/GPU:0", "/XLA_GPU:1"])

I've tried all of the above, but it doesn't work. If you know how to fix, please answer. You cannot reduce the image size. (This is the purpose of my experiment) And it says that the GPU memory is 32GB, but if you let me know why the error occurs when it is not 32GB, I would really appreciate it.

...

ANSWER

Answered 2020-Oct-04 at 05:42

Short answer. Stop using Flatten and change it to GlobalAveragePooling2D or GlobalMaxPooling2D.

Long answer. What you're trying is use 64*64*800 as input size of a Fully connected layer which is wayyyy too big. Modern CNN models are all no longer use Flatten and replace it with GlobalAveragePooling2D as it also reserve variable size capability of CNN.

Source https://stackoverflow.com/questions/64191241

QUESTION

Im not running anything on GPU. But GPU stat is not clearing my usage

Asked 2020-Jul-14 at 05:00

GPUstat image

Im not running anything on gpu. My ID is suil5044. GPUstat is not clearing my usage. I think I finished it in the IDE but actually the code is still running on the server.

How do I kill my code still running on the server? And not affect other users

Thanks!

...

ANSWER

Answered 2020-Jul-14 at 05:00

It would be better if you could show us which process are using your GPU. Because sometimes many utils process use the GPU too under same user name.

However, please try,

Find out all running process pid which are using the GPU. This tool may help.
Kill process by PID by this command: kill -9 (be sure this pid is under your username)

Source https://stackoverflow.com/questions/62888242

QUESTION

Why can't I run tensorflow session on CPU while one GPU device's memory is all allocated?

Asked 2019-Dec-29 at 23:15

From the tensorflow website (https://www.tensorflow.org/guide/using_gpu) I found the following code to manually specify the use of a CPU instead of a GPU:

...

ANSWER

Answered 2019-Dec-29 at 23:15

OK... so with the help of my colleague, I have a workable solution. The key is, in fact, a modification to the config. Specifically, something like this:

Source https://stackoverflow.com/questions/53157197

QUESTION

Is it possible to train a H2O model with GPU and predict with a CPU?

Asked 2019-Oct-23 at 06:00

For trainining speed, it would be nice to be able to train a H2O model with GPUs, take the model file, and then predict on a machine without GPUs.

It seems like that should be possible in theory, but with the H2O release 3.13.0.341, that doesn't seem to happen, except for XGBoost model.

When I run gpustat -cup I can see the GPUs kick in when I train H2O's XGBoost model. This doesn't happen with DL, DRF, GLM, or GBM.

I wouldn't be surprised if a difference in float point size (16, 32, 64) could cause some inconsistency, not to mention the vagaries due to multiprocessor modeling, but I think I could live with that.

(This is related to my question here, but now that I understand the environment better I can see that the GPUs aren't used all the time.)

How can I tell if H2O 3.11.0.266 is running with GPUs?

...

ANSWER

Answered 2017-Jul-26 at 04:13

The new XGBoost integration in H2O is the only GPU-capable algorithm in H2O (proper) at this time. So you can train an XGBoost model on GPUs and score on CPUs, but that's not true for the other H2O algorithms.

There is also the H2O Deep Water project, which provides integration between H2O and three third-party deep learning backends (MXNet, Caffe and TensorFlow), all of which are GPU-capable. So you can train those models using a GPU and score on a CPU as well. You can download the H2O Deep Water jar file (or R package, or Python module) at the Deep Water link above, and you can find out more info in the Deep Water GitHub repo README.

Source https://stackoverflow.com/questions/45290052

QUESTION

OpenGL program with tensorflow C++ gives failed call to cuInit : CUDA_ERROR_OUT_OF_MEMORY

Asked 2018-Jan-10 at 15:04

I have trained a model with no issues using tensorflow on python. I am now trying to integrate inference for this model into a pre-existing OpenGL enabled software. However, I get a CUDA_ERROR_OUT_OF_MEMORY during cuInit (that is, even earlier than loading the model, just at session creation). It does seem, that OpenGL has taken some MiBs of memory (around 300 MB), as shown by gpustat or nvidia-smi.

Is it possible there is a clash as both TF and OpenGL are trying to access/allocate the GPU memory? Has anyone encountered this problem before? Most references I found googling around are at model loading time, not at session/CUDA initialization. Is this completely unrelated to OpenGL and I am just barking up the wrong tree? A simple TF C++ inference example works. Any help is appreciated.

Here is the tensorflow logging output, for completeness:

...

ANSWER

Answered 2018-Jan-10 at 15:02

Ok, the problem was the use of the sanitizer in the debug version of the binary. The release version, or the debug version with no sanitizer work as expected.

Source https://stackoverflow.com/questions/48154126

Community Discussions, Code Snippets contain sources that include Stack Exchange Network

Vulnerabilities

No vulnerabilities reported

Install gpustat

If you don’t have root privilege, please try to install on user namespace: pip install --user gpustat.

Support

For any new features, suggestions and bugs create an issue on GitHub. If you have any questions check and ask questions on community page Stack Overflow .

Find more information at: