Metrics de segmentation

Deux metrics seront utilisés pour testé le modèle: Intersection-Over-Union (IoU, Jaccard Index) and the F1 score (Dice coefficient) Références:

Intersection-Over-Union (IoU, Jaccard Index)

Références:

https://scikit-learn.org/stable/modules/generated/sklearn.metrics.jaccard_score.html

F1 score (Dice coefficient)

Références:

https://scikit-learn.org/stable/modules/generated/sklearn.metrics.f1_score.html

Premier script de test

Python 3.6 virtual environment

Python env 3.6

Python 3.6
Création de l’environnement Python virtuel
Installation des modules Python requis

salloc --account=def-germ2201-ab --ntasks=1 --nodes=1 --cpus-per-task=1 --mem=4800M --time=0-3:00

module load python/3.6

# clean the python virtual pyenv36
rm -rf $SLURM_TMPDIR/pyenv36

# create the virtual env
virtualenv --no-download $SLURM_TMPDIR/pyenv36

# activate the python virtual env
source $SLURM_TMPDIR/pyenv36/bin/activate

# check the python and pip version (3.6)
python --version
pip --version

# install requirements
pip3 install --no-index torch
pip3 install --no-index scikit-learn
pip3 install --no-index scikit-image
pip3 install --no-index keras
pip3 install --no-index tensorflow_gpu
pip3 install --no-index jupyter

Préparation des images

from skimage import io
import numpy as np
import os

filename = os.path.join('/home/vincelf/DeepLabv3FineTuning-master/CrackForest/Images', '001.jpg')
im = io.imread(filename)
im.shape
# (320, 480, 3)
im2 = np.expand_dims(im,0)
im2.shape
# (1, 320, 480, 3)

filename = os.path.join('/home/vincelf/DeepLabv3FineTuning-master/CrackForest/Masks', '001_label.PNG')
mask = io.imread(filename)
mask.shape
# (320, 480)
m = np.array(mask)
m2 = np.stack((m,)*3,-1)
m2.shape
# (320, 480, 3)
m3 = np.expand_dims(m2,0)
m3.shape
# (1, 320, 480, 3)

Test avec des images du dataset de deepscene

from skimage import io
import numpy as np
import os

home='/home/vincelf/downloads/freiburg_forest_annotated'
rgb='train/rgb/b102-798_Clipped.jpg'
gt_color='train/GT_color/b102-798_mask.png'

filename = os.path.join(home, rgb)
im = io.imread(filename)
im.shape

filename = os.path.join(home, gt_color)
mask = io.imread(filename)
mask.shape

def iou_score(y_true, y_pred):
  intersection = np.logical_and(y_true, y_pred)
  union = np.logical_or(y_true, y_pred)
  iou_score = np.sum(intersection) / np.sum(union)
  return iou_score

iou_score(im,mask)

# Both y_true and y_pred are m x r x c x n where m is the number of images, r is the number of rows, c is the number of columns, and n is the number of classes.
from keras import backend as K
def iou_coef(y_true, y_pred, smooth=1):
  intersection = K.sum(K.abs(y_true * y_pred), axis=[1,2,3])
  union = K.sum(y_true,[1,2,3])+K.sum(y_pred,[1,2,3])-intersection
  iou = K.mean((intersection + smooth) / (union + smooth), axis=0)
  return iou
  
# error returned with tensorflow 1.15.0 et 2.2

iou_coef(im, mask)

from skimage import io
import numpy as np
import os

home_gt='/home/vincelf/downloads/freiburg_forest_annotated/test/GT_color'
gt_color='b1-09517_Clipped.png'
gt_color_f = os.path.join(home_gt, gt_color)
gt_color_img = io.imread(gt_color_f)
gt_color_img.shape
set( tuple(v) for m2d in gt_color_img for v in m2d )

home_pred='/home/vincelf/upload'
color_pred='b1-09517_Clipped_pred_new_color.jpg'
color_pred_f = os.path.join(home_pred, color_pred)
color_pred_img = io.imread(color_pred_f)
color_pred_img.shape
set( tuple(v) for m2d in color_pred_img for v in m2d )

Test 1st implementation

Reference: https://www.jeremyjordan.me/evaluating-image-segmentation-models/

def iou_score(y_true, y_pred):
  intersection = np.logical_and(y_true, y_pred)
  union = np.logical_or(y_true, y_pred)
  iou_score = np.sum(intersection) / np.sum(union)
  return iou_score

iou_score(im2,m3)
# 0.011920572916666667

Test 2nd implementation

Reference: https://towardsdatascience.com/metrics-to-evaluate-your-semantic-segmentation-model-6bcb99639aa2

Retourne une erreur

# Both y_true and y_pred are m x r x c x n where m is the number of images, r is the number of rows, c is the number of columns, and n is the number of classes.
from keras import backend as K
def iou_coef(y_true, y_pred, smooth=1):
  intersection = K.sum(K.abs(y_true * y_pred), axis=[1,2,3])
  union = K.sum(y_true,[1,2,3])+K.sum(y_pred,[1,2,3])-intersection
  iou = K.mean((intersection + smooth) / (union + smooth), axis=0)
  return iou
  
iou_coef(im2, m3)

Erreur:

2020-06-05 00:48:26.050960: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2020-06-05 00:48:26.096634: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: 
pciBusID: 0000:1e:00.0 name: Tesla V100-SXM2-16GB computeCapability: 7.0
coreClock: 1.53GHz coreCount: 80 deviceMemorySize: 15.78GiB deviceMemoryBandwidth: 836.37GiB/s
2020-06-05 00:48:26.097152: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcudart.so.10.1'; dlerror: libcudart.so.10.1: cannot open shared object file: No such file or directory
2020-06-05 00:48:26.097521: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcublas.so.10'; dlerror: libcublas.so.10: cannot open shared object file: No such file or directory
2020-06-05 00:48:26.097727: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcufft.so.10'; dlerror: libcufft.so.10: cannot open shared object file: No such file or directory
2020-06-05 00:48:26.097878: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcurand.so.10'; dlerror: libcurand.so.10: cannot open shared object file: No such file or directory
2020-06-05 00:48:26.098051: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcusolver.so.10'; dlerror: libcusolver.so.10: cannot open shared object file: No such file or directory
2020-06-05 00:48:26.098187: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcusparse.so.10'; dlerror: libcusparse.so.10: cannot open shared object file: No such file or directory
2020-06-05 00:48:26.098325: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcudnn.so.7'; dlerror: libcudnn.so.7: cannot open shared object file: No such file or directory
2020-06-05 00:48:26.098334: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1598] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.
Skipping registering GPU devices...
2020-06-05 00:48:26.099037: I tensorflow/core/platform/cpu_feature_guard.cc:143] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 AVX512F FMA
2020-06-05 00:48:26.110382: I tensorflow/core/platform/profile_utils/cpu_utils.cc:102] CPU Frequency: 2400000000 Hz
2020-06-05 00:48:26.110945: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x45c9ce0 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2020-06-05 00:48:26.110963: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version
2020-06-05 00:48:26.113305: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-06-05 00:48:26.113319: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1108]      
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<stdin>", line 2, in iou_coef
  File "/localscratch/vincelf.8491172.0/pyenv36/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py", line 1900, in abs
    return tf.abs(x)
  File "/localscratch/vincelf.8491172.0/pyenv36/lib/python3.6/site-packages/tensorflow/python/util/dispatch.py", line 180, in wrapper
    return target(*args, **kwargs)
  File "/localscratch/vincelf.8491172.0/pyenv36/lib/python3.6/site-packages/tensorflow/python/ops/math_ops.py", line 268, in abs
    return gen_math_ops._abs(x, name=name)
  File "/localscratch/vincelf.8491172.0/pyenv36/lib/python3.6/site-packages/tensorflow/python/ops/gen_math_ops.py", line 52, in _abs
    _ops.raise_from_not_ok_status(e, name)
  File "/localscratch/vincelf.8491172.0/pyenv36/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 6653, in raise_from_not_ok_status
    six.raise_from(core._status_to_exception(e.code, message), None)
  File "<string>", line 3, in raise_from
tensorflow.python.framework.errors_impl.NotFoundError: Could not find valid device for node.
Node:
All kernels registered for op Abs :
  device='XLA_GPU'; T in [DT_FLOAT, DT_DOUBLE, DT_INT32, DT_INT16, DT_INT8, DT_INT64, DT_BFLOAT16, DT_HALF]
  device='XLA_CPU'; T in [DT_FLOAT, DT_DOUBLE, DT_INT32, DT_INT16, DT_INT8, DT_INT64, DT_BFLOAT16, DT_HALF]
  device='XLA_CPU_JIT'; T in [DT_FLOAT, DT_DOUBLE, DT_INT32, DT_INT16, DT_INT8, DT_INT64, DT_BFLOAT16, DT_HALF]
  device='XLA_GPU_JIT'; T in [DT_FLOAT, DT_DOUBLE, DT_INT32, DT_INT16, DT_INT8, DT_INT64, DT_BFLOAT16, DT_HALF]
  device='GPU'; T in [DT_INT32]
  device='GPU'; T in [DT_INT64]
  device='GPU'; T in [DT_DOUBLE]
  device='GPU'; T in [DT_FLOAT]
  device='GPU'; T in [DT_HALF]
  device='CPU'; T in [DT_INT64]
  device='CPU'; T in [DT_INT32]
  device='CPU'; T in [DT_INT16]
  device='CPU'; T in [DT_INT8]
  device='CPU'; T in [DT_DOUBLE]
  device='CPU'; T in [DT_FLOAT]
  device='CPU'; T in [DT_BFLOAT16]
  device='CPU'; T in [DT_HALF]
 [Op:Abs]

Test 3rd implementation

Reference: https://scikit-learn.org/stable/modules/generated/sklearn.metrics.jaccard_score.html

Retourne une erreur

from sklearn.metrics import jaccard_score
jaccard_score(im2, m3, average='samples')

Erreur:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/localscratch/vincelf.8491172.0/pyenv36/lib/python3.6/site-packages/sklearn/utils/validation.py", line 73, in inner_f
    return f(**kwargs)
  File "/localscratch/vincelf.8491172.0/pyenv36/lib/python3.6/site-packages/sklearn/metrics/_classification.py", line 736, in jaccard_score
    pos_label)
  File "/localscratch/vincelf.8491172.0/pyenv36/lib/python3.6/site-packages/sklearn/metrics/_classification.py", line 1250, in _check_set_wise_labels
    y_type, y_true, y_pred = _check_targets(y_true, y_pred)
  File "/localscratch/vincelf.8491172.0/pyenv36/lib/python3.6/site-packages/sklearn/metrics/_classification.py", line 98, in _check_targets
    raise ValueError("{0} is not supported".format(y_type))
ValueError: unknown is not supported

Using open source

Using tensor flow implementation

https://github.com/tensorflow/tensorflow/blob/r1.13/tensorflow/python/ops/metrics_impl.py#L1073

Using matlab

Références:

dynimc allocation of one CPU to test

One cpu the test script will work; to use multiple CPU, it’s another matlab command

salloc --account=def-germ2201-ab --ntasks=1 --nodes=1 --cpus-per-task=1 --mem=4800M --time=0-3:00

Load the matlab module

module load matlab/2018a

Test script

Référence: https://docs.computecanada.ca/wiki/MATLAB

function cosplot()
% MATLAB file example to approximate a sawtooth
% with a truncated Fourier expansion.
nterms=5;
fourbypi=4.0/pi;
np=100;
y(1:np)=pi/2.0;
x(1:np)=linspace(-2.0*pi,2*pi,np);

for k=1:nterms
 twokm=2*k-1;
 y=y-fourbypi*cos(twokm*x)/twokm^2;
end

plot(x,y)
print -dpsc matlab_test_plot.ps
quit
end

[vincelf@blg5608 matlab]$ salloc --account=def-germ2201-ab --ntasks=1 --nodes=1 --cpus-per-task=1 --mem=4800M --time=0-3:00
salloc: Pending job allocation 8566547
salloc: job 8566547 queued and waiting for resources
salloc: job 8566547 has been allocated resources
salloc: Granted job allocation 8566547
salloc: Waiting for resource configuration
salloc: Nodes blg8610 are ready for job
[vincelf@blg8610 matlab]$ module load matlab/2018a
[vincelf@blg8610 matlab]$ pwd
/home/vincelf/matlab
[vincelf@blg8610 matlab]$ srun matlab -nodisplay -singleCompThread -r "test"
Opening log file:  /tmp/java.log.28720

                            < M A T L A B (R) >
                  Copyright 1984-2018 The MathWorks, Inc.
                   R2018a (9.4.0.813654) 64-bit (glnxa64)
                             February 23, 2018


 
To get started, type one of these: helpwin, helpdesk, or demo.
For product information, visit www.mathworks.com.
 
[vincelf@blg8610 matlab]$ squeue -u vincelf
[vincelf@blg8610 matlab]$ seff <jobid> 
[vincelf@blg8610 matlab]$ scancel -u vincelf <jobid> 
[vincelf@blg8610 matlab]$ ls -al /tmp (pour la job java)

Errors

Matlab: Invalid use of operator

When launching script with absolute path on the command line. launch script with relative path.

[vincelf@blg4121 matlab]$ srun matlab -nodisplay -singleCompThread -r "/home/vincelf/matlab/test"
Opening log file:  /tmp/java.log.16513

                            < M A T L A B (R) >
                  Copyright 1984-2018 The MathWorks, Inc.
                   R2018a (9.4.0.813654) 64-bit (glnxa64)
                             February 23, 2018

 
To get started, type one of these: helpwin, helpdesk, or demo.
For product information, visit www.mathworks.com.
 
 /home/vincelf/matlab/test
 |
Error: Invalid use of operator.
 





^Csrun: interrupt (one more within 1 sec to abort)
srun: step:8570507.7 task 0: running


^Csrun: interrupt (one more within 1 sec to abort)
srun: step:8570507.7 task 0: running
^Csrun: interrupt (one more within 1 sec to abort)
srun: step:8570507.7 task 0: running
^Csrun: sending Ctrl-C to job 8570507.7
>> >> >> >> >> >> >> >> 
srun: Job step aborted: Waiting up to 62 seconds for job step to finish.
slurmstepd: error: *** STEP 8570507.7 ON blg4121 CANCELLED AT 2020-06-07T04:14:12 ***
>> [vincelf@blg4121 matlab]$ 
[vincelf@blg4121 matlab]$ 
[vincelf@blg4121 matlab]$ squeue -u vincelf
          JOBID     USER      ACCOUNT           NAME  ST  TIME_LEFT NODES CPUS TRES_PER_N MIN_MEM NODELIST (REASON) 
        8570507  vincelf def-germ2201             sh   R    2:51:01     1    1        N/A   4800M blg4121 (None) 
[vincelf@blg4121 matlab]$ scancel -u vincelf 8570507
salloc: Job allocation 8570507 has been revoked.
srun: Job step aborted: Waiting up to 62 seconds for job step to finish.
[vincelf@blg4121 matlab]$ exit
[vincelf@beluga3 matlab]$ salloc --account=def-germ2201-ab --ntasks=1 --nodes=1 --cpus-per-task=1 --mem=4800M --time=0-3:00
salloc: Pending job allocation 8571292
salloc: job 8571292 queued and waiting for resources
salloc: job 8571292 has been allocated resources
salloc: Granted job allocation 8571292
salloc: Waiting for resource configuration
salloc: Nodes blg8339 are ready for job
[vincelf@blg8339 matlab]$ module load matlab/2018a
[vincelf@blg8339 matlab]$ srun matlab -nodisplay -singleCompThread -r "test"
Opening log file:  /tmp/java.log.36150

                            < M A T L A B (R) >
                  Copyright 1984-2018 The MathWorks, Inc.
                   R2018a (9.4.0.813654) 64-bit (glnxa64)
                             February 23, 2018

 
To get started, type one of these: helpwin, helpdesk, or demo.
For product information, visit www.mathworks.com.
 
[vincelf@blg8339 matlab]$ 

Autres métriques

NVIDIA tegrastats, pour les stats mem, cpu, gpu, chaleur, power

Note

AO: always on,
PLL: phase locking loop,
PMIC: power management IC,
Tdiode: temperature of thermal diode in chip,
Tboard: temperature of board read from a thermal diode on board. Reference: https://forums.developer.nvidia.com/t/tx1-thermal-sensor-location/43348/2

Les indicateurs matériels suivants sont disponibles sur le Jetson Nano: Référence: https://docs.nvidia.com/jetson/l4t/index.html#page/Tegra%2520Linux%2520Driver%2520Package%2520Development%2520Guide%2FAppendixTegraStats.html%23wwpID0E0SB0HA

Exemple RAM 2072/3956MB (lfb 1x2MB) SWAP 893/1978MB (cached 35MB) IRAM 0/252kB(lfb 252kB) CPU [3%@1479,2%@1479,3%@1479,2%@1479] EMC_FREQ 3%@1600 GR3D_FREQ 0%@921 APE 25 PLL@22.5C CPU@24.5C PMIC@100C GPU@24C AO@31C thermal@24C POM_5V_IN 2159/3257 POM_5V_GPU 124/557 POM_5V_CPU 332/657

Exécution: Une seul fois dans un fichier:

$ tegrastats | while read a; do echo "$a" | awk -v now="$(date +%s)" '{OFS=" "}{print now" "$0}' > tegrastats-`date +%Y%m%dT%H%M%S`.log; tegrastats --stop; done

en mode continue jusqu’au ctrl-c

FILENAME=tegrastats-`date +%Y%m%dT%H%M%S`.log; sudo tegrastats | while read a; do echo "$a" | awk -v now="$(date +%s)" '{OFS=" "}{print now" "$0}' >> ${FILENAME};  done
# ou 
sudo tegrastats > tegrastats.log
ctrl-c

Exemple:

RAM 3300/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [18%@1479,13%@1479,15%@1479,26%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 96%@921 APE 25 PLL@32C CPU@35C PMIC@100C GPU@34C AO@42.5C thermal@34C POM_5V_IN 6289/5162 POM_5V_GPU 2495/1845 POM_5V_CPU 1006/743
RAM 3300/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [20%@1479,13%@1479,15%@1479,11%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 96%@921 APE 25 PLL@32C CPU@35C PMIC@100C GPU@33.5C AO@42.5C thermal@33.75C POM_5V_IN 6057/5177 POM_5V_GPU 2463/1856 POM_5V_CPU 927/746
RAM 3300/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [24%@1479,9%@1479,20%@1479,13%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 97%@921 APE 25 PLL@31.5C CPU@34.5C PMIC@100C GPU@34C AO@42.5C thermal@34C POM_5V_IN 5550/5183 POM_5V_GPU 2309/1863 POM_5V_CPU 567/743
RAM 3300/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [13%@1479,14%@1479,14%@1479,26%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 96%@921 APE 25 PLL@31.5C CPU@35C PMIC@100C GPU@34C AO@42.5C thermal@34C POM_5V_IN 5550/5189 POM_5V_GPU 2309/1870 POM_5V_CPU 567/740
RAM 3300/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [13%@1479,13%@1479,16%@1479,22%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 99%@921 APE 25 PLL@32C CPU@34.5C PMIC@100C GPU@33.5C AO@42.5C thermal@34C POM_5V_IN 5356/5192 POM_5V_GPU 2150/1875 POM_5V_CPU 568/737
RAM 3301/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [13%@1479,11%@1479,15%@1479,26%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 99%@921 APE 25 PLL@31.5C CPU@34.5C PMIC@100C GPU@34C AO@42.5C thermal@34C POM_5V_IN 5242/5192 POM_5V_GPU 2028/1877 POM_5V_CPU 568/735
RAM 3300/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [15%@1479,12%@1479,14%@1479,23%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 99%@921 APE 25 PLL@32C CPU@34C PMIC@100C GPU@33.5C AO@42.5C thermal@34.25C POM_5V_IN 4998/5189 POM_5V_GPU 1788/1876 POM_5V_CPU 650/733
RAM 3301/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [13%@1479,12%@1479,7%@1479,24%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 99%@921 APE 25 PLL@32C CPU@35C PMIC@100C GPU@34C AO@42.5C thermal@34.5C POM_5V_IN 5315/5191 POM_5V_GPU 1744/1874 POM_5V_CPU 852/735
RAM 3300/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [13%@1479,14%@1479,15%@1479,14%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 99%@921 APE 25 PLL@32C CPU@35C PMIC@100C GPU@34C AO@42.5C thermal@34.5C POM_5V_IN 5437/5195 POM_5V_GPU 1742/1872 POM_5V_CPU 972/739
RAM 3300/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [11%@1479,17%@1479,23%@1479,10%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 99%@921 APE 25 PLL@32C CPU@34.5C PMIC@100C GPU@34C AO@42.5C thermal@34.75C POM_5V_IN 5080/5193 POM_5V_GPU 1788/1871 POM_5V_CPU 650/738
RAM 3301/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [26%@1479,15%@1479,8%@1479,10%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 96%@921 APE 25 PLL@32C CPU@34.5C PMIC@100C GPU@34C AO@43C thermal@34.75C POM_5V_IN 5080/5192 POM_5V_GPU 1828/1870 POM_5V_CPU 568/735
RAM 3301/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [29%@1479,17%@1479,11%@1479,12%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 98%@921 APE 25 PLL@32C CPU@34.5C PMIC@100C GPU@34C AO@42.5C thermal@34.5C POM_5V_IN 5356/5194 POM_5V_GPU 2025/1872 POM_5V_CPU 607/733
RAM 3301/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [28%@1479,13%@1479,7%@1479,10%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 99%@921 APE 25 PLL@32C CPU@34.5C PMIC@100C GPU@34C AO@43C thermal@34.5C POM_5V_IN 5662/5201 POM_5V_GPU 2265/1878 POM_5V_CPU 606/731
RAM 3301/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [26%@1479,14%@1479,9%@1479,7%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 99%@921 APE 25 PLL@32C CPU@35C PMIC@100C GPU@34C AO@42.5C thermal@34.5C POM_5V_IN 5896/5210 POM_5V_GPU 2544/1887 POM_5V_CPU 565/729
RAM 3301/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [30%@1479,12%@1479,11%@1479,13%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 92%@921 APE 25 PLL@32C CPU@35C PMIC@100C GPU@34C AO@43C thermal@34.5C POM_5V_IN 5743/5218 POM_5V_GPU 2022/1889 POM_5V_CPU 1011/733
RAM 3301/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [12%@1479,12%@1479,11%@1479,11%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 99%@921 APE 25 PLL@32C CPU@34.5C PMIC@100C GPU@34C AO@43C thermal@34.5C POM_5V_IN 6057/5229 POM_5V_GPU 2580/1899 POM_5V_CPU 766/733
RAM 3304/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [16%@1479,13%@1479,27%@1479,10%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 99%@921 APE 25 PLL@32C CPU@34.5C PMIC@100C GPU@34C AO@43C thermal@35.25C POM_5V_IN 6128/5241 POM_5V_GPU 2741/1910 POM_5V_CPU 564/731
RAM 3304/3956MB (lfb 5x2MB) SWAP 1341/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [13%@1479,12%@1479,13%@1479,25%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 99%@921 APE 25 PLL@32C CPU@35C PMIC@100C GPU@34.5C AO@43.5C thermal@34.5C POM_5V_IN 6168/5254 POM_5V_GPU 2701/1920 POM_5V_CPU 604/729
RAM 3308/3956MB (lfb 5x2MB) SWAP 1355/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [29%@1479,13%@1479,20%@1479,33%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 99%@921 APE 25 PLL@32.5C CPU@35C PMIC@100C GPU@34.5C AO@43C thermal@34.75C POM_5V_IN 6671/5272 POM_5V_GPU 2808/1932 POM_5V_CPU 882/731
RAM 3310/3956MB (lfb 4x2MB) SWAP 1355/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [25%@1479,12%@1479,23%@1479,28%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 96%@921 APE 25 PLL@32C CPU@35C PMIC@100C GPU@34.5C AO@43C thermal@34.75C POM_5V_IN 5428/5274 POM_5V_GPU 2228/1936 POM_5V_CPU 567/729
RAM 3307/3956MB (lfb 4x2MB) SWAP 1355/1978MB (cached 5MB) IRAM 0/252kB(lfb 252kB) CPU [6%@1479,12%@1479,14%@1479,26%@1479] EMC_FREQ 29%@1600 GR3D_FREQ 99%@921 APE 25 PLL@32C CPU@35.5C PMIC@100C GPU@34.5C AO@43C thermal@34.75C POM_5V_IN 5824/5282 POM_5V_GPU 2503/1943 POM_5V_CPU 605/728

free -m, pour la mémoire utilisée, cache, swap et libre

Pour garder à l’écran :

$ watch -n 5 free -m
ctrl-c

Pour logguer successivement:

# sudo free -m -s 1 -t -l > free-`date +%Y%m%dT%H%M%S`.log
FILENAME=free-`date +%Y%m%dT%H%M%S`.log; sudo free -m -s 1 -t -l | while read a; do echo "$a" | awk -v now="$(date +%s)" '{OFS=" "}{if(NF==7) {print now" "$0;} else {print $0;}}' >> ${FILENAME};  done
ctrl-c

Exemple:

total        used        free      shared  buff/cache   available
1594438559 Mem:           3956        2377         535         124        1044        1572
Low:           3956        3421         535
High:             0           0           0
Swap:          1978         388        1590
Total:         5934        2765        2125

total        used        free      shared  buff/cache   available
1594438560 Mem:           3956        2377         534         124        1044        1571
Low:           3956        3422         534
High:             0           0           0
Swap:          1978         388        1590
Total:         5934        2765        2124

total        used        free      shared  buff/cache   available
1594438561 Mem:           3956        2380         531         124        1044        1568
Low:           3956        3424         531
High:             0           0           0
Swap:          1978         388        1590
Total:         5934        2768        2121

total        used        free      shared  buff/cache   available
1594438562 Mem:           3956        2376         535         124        1044        1572
Low:           3956        3420         535
High:             0           0           0
Swap:          1978         388        1590
Total:         5934        2764        2125

iotop, pour le IO

Référence: https://www.opsdash.com/blog/disk-monitoring-linux.html

Installe:

$ sudo apt-get install iotop

Exécute en mode continue jusqu’au ctrl-c:

$ FILENAME="iotop-`date +%Y%m%dT%H%M%S`.log"; sudo iotop --batch --time --only --processes --kilobytes > $FILENAME
ctrl-c

Extrait les infos:
segnet-camera

$ grep -E "segnet" iotop.log
57:55 17245 ?sys lefv2603 12849.44 K/s    0.00 K/s  0.00 %  9.61 % segnet-camera --network=fcn-resnet18-deepscene
57:56 17245 ?sys lefv2603 29691.41 K/s    0.00 K/s  0.00 % 33.52 % segnet-camera --network=fcn-resnet18-deepscene
57:57 17245 ?sys lefv2603 24320.22 K/s    0.00 K/s  0.00 % 27.34 % segnet-camera --network=fcn-resnet18-deepscene
57:58 17245 ?sys lefv2603 30579.35 K/s    0.00 K/s  0.00 % 37.04 % segnet-camera --network=fcn-resnet18-deepscene
57:59 17245 ?sys lefv2603 23505.52 K/s    0.00 K/s  0.00 % 25.05 % segnet-camera --network=fcn-resnet18-deepscene
58:00 17245 ?sys lefv2603 30970.10 K/s    0.00 K/s  0.00 % 36.01 % segnet-camera --network=fcn-resnet18-deepscene
58:02 17245 ?sys lefv2603 25219.86 K/s    0.00 K/s  0.00 % 31.23 % segnet-camera --network=fcn-resnet18-deepscene
58:03 17245 ?sys lefv2603 27602.30 K/s    0.00 K/s  0.00 % 33.10 % segnet-camera --network=fcn-resnet18-deepscene
58:04 17245 ?sys lefv2603 16928.24 K/s    6.90 K/s  0.00 %  7.22 % segnet-camera --network=fcn-resnet18-deepscene
58:06 17245 ?sys lefv2603 16325.93 K/s    0.00 K/s  0.00 %  3.96 % segnet-camera --network=fcn-resnet18-deepscene
58:08 17245 ?sys lefv2603 5186.13 K/s    3.16 K/s  0.00 %  1.85 % segnet-camera --network=fcn-resnet18-deepscene
58:09 17245 ?sys lefv2603    0.00 K/s    6.50 K/s  0.00 %  0.00 % segnet-camera --network=fcn-resnet18-deepscene
58:10 17245 ?sys lefv2603    0.00 K/s    3.36 K/s  0.00 %  0.00 % segnet-camera --network=fcn-resnet18-deepscene
58:11 17245 ?sys lefv2603    0.00 K/s    6.78 K/s  0.00 %  0.00 % segnet-camera --network=fcn-resnet18-deepscene
58:12 17245 ?sys lefv2603    0.00 K/s    6.78 K/s  0.00 %  0.00 % segnet-camera --network=fcn-resnet18-deepscene
58:14 17245 ?sys lefv2603    0.00 K/s    3.38 K/s  0.00 %  0.00 % segnet-camera --network=fcn-resnet18-deepscene
58:15 17245 ?sys lefv2603    0.00 K/s    6.78 K/s  0.00 %  0.00 % segnet-camera --network=fcn-resnet18-deepscene
58:16 17245 ?sys lefv2603    0.00 K/s    3.39 K/s  0.00 %  0.00 % segnet-camera --network=fcn-resnet18-deepscene
58:17 17245 ?sys lefv2603 1550.39 K/s    6.80 K/s  0.00 %  0.12 % segnet-camera --network=fcn-resnet18-deepscene

DISK

$ grep -E "Total DISK" iotop.log
57:55 Total DISK READ :   12849.44 K/s | Total DISK WRITE :       0.00 K/s
57:56 Total DISK READ :   29691.41 K/s | Total DISK WRITE :       3.48 K/s
57:57 Total DISK READ :   24320.22 K/s | Total DISK WRITE :       0.00 K/s
57:58 Total DISK READ :   30579.35 K/s | Total DISK WRITE :       0.00 K/s
57:59 Total DISK READ :   23505.52 K/s | Total DISK WRITE :      10.46 K/s
58:00 Total DISK READ :   30970.10 K/s | Total DISK WRITE :       3.46 K/s
58:02 Total DISK READ :   25219.86 K/s | Total DISK WRITE :      13.90 K/s
58:03 Total DISK READ :   27602.30 K/s | Total DISK WRITE :     425.98 K/s
58:04 Total DISK READ :   22860.72 K/s | Total DISK WRITE :      17.25 K/s
58:05 Total DISK READ :    1552.91 K/s | Total DISK WRITE :    1143.16 K/s
58:06 Total DISK READ :   18201.47 K/s | Total DISK WRITE :      19.44 K/s
58:08 Total DISK READ :   28000.67 K/s | Total DISK WRITE :      31.60 K/s
58:09 Total DISK READ :   12615.58 K/s | Total DISK WRITE :      97.54 K/s
58:10 Total DISK READ :    9141.37 K/s | Total DISK WRITE :       3.36 K/s
58:11 Total DISK READ :   38425.82 K/s | Total DISK WRITE :      10.17 K/s
58:12 Total DISK READ :   37416.45 K/s | Total DISK WRITE :      13.56 K/s
58:14 Total DISK READ :   38507.38 K/s | Total DISK WRITE :     450.08 K/s

dstat, pour les stats system, cpu, mem

Installe:

sudo apt-get install dstat

Exécute jusqu’au ctrl-c

$ sudo dstat -gmrsty -c -C 0,1,2,3,total --output dstat-`date +%Y%m%dT%H%M%S`.log
ctrl-c

Exemple:

"Dstat 0.7.3 CSV output"
"Author:","Dag Wieers <dag@wieers.com>",,,,"URL:","http://dag.wieers.com/home-made/dstat/"
"Host:","lefv2603-desktop",,,,"User:","lefv2603"
"Cmdline:","dstat -gmrsty -c -C 0,1,2,3,total --output dstat.log",,,,"Date:","08 Jul 2020 23:10:25 EDT"
"paging",,"memory usage",,,,"io/total",,"swap",,"system","system",,"cpu0 usage",,,,,"cpu1 usage",,,,,"cpu2 usage",,,,,"cpu3 usage",,,,,"total cpu usage",,,,
"in","out","used","free","buff","cach","read","writ","used","free","time","int","csw","cpu0 usage:usr","cpu0 usage:sys","cpu0 usage:idl","cpu0 usage:wai","cpu0 usage:stl","cpu1 usage:usr","cpu1 usage:sys","cpu1 usage:idl","cpu1 usage:wai","cpu1 usage:stl","cpu2 usage:usr","cpu2 usage:sys","cpu2 usage:idl","cpu2 usage:wai","cpu2 usage:stl","cpu3 usage:usr","cpu3 usage:sys","cpu3 usage:idl","cpu3 usage:wai","cpu3 usage:stl","total cpu usage:usr","total cpu usage:sys","total cpu usage:idl","total cpu usage:wai","total cpu usage:stl"
37899.432,167917.961,2784907264,924696576,10518528,572469248,22.964,46.010,1407582208,666632192,08-07 23:10:25,2476.053,4135.282,6.071,2.780,88.381,2.768,0,5.846,2.723,88.560,2.871,0,5.897,2.763,88.489,2.851,0,5.846,2.374,89.451,2.329,0,5.915,2.660,88.720,2.705,0
266240,0,2784669696,921223168,10526720,576299008,105,0,1407315968,666898432,08-07 23:10:26,5758,7913,2.970,0.990,92.079,3.960,0,1.010,2.020,96.970,0,0,3.030,3.030,90.909,3.030,0,2,6,92,0,0,2.015,2.771,93.451,1.763,0
151552,0,2784800768,921157632,10526720,576106496,37,0,1407164416,667049984,08-07 23:10:27,5387,8001,2,1,97,0,0,0,1.020,98.980,0,0,0,5,95,0,0,3.030,2.020,94.949,0,0,1.754,2.256,95.990,0,0
0,0,2781556736,924401664,10526720,576106496,0,0,1407164416,667049984,08-07 23:10:28,5569,8296,3.030,1.010,95.960,0,0,0,2.020,97.980,0,0,0.990,3.960,95.050,0,0,1,5,94,0,0,1.250,3,95.750,0,0
0,0,2781515776,924434432,10534912,576106496,0,4,1407164416,667049984,08-07 23:10:29,5661,8362,0,1.010,98.990,0,0,2.020,5.051,92.929,0,0,2,4,93,1,0,0,2.970,97.030,0,0,1.003,3.509,95.489,0,0
0,0,2781581312,924368896,10534912,576106496,0,0,1407164416,667049984,08-07 23:10:30,5769,8511,1.980,1.980,96.040,0,0,0,2.020,97.980,0,0,2,4,94,0,0,4,2,94,0,0,1.754,2.506,95.739,0,0

Combiné pendant l’inférence avec la caméra

$ sudo tegrastats > tegrastats-`date +%Y%m%dT%H%M%S`.log | sudo iotop --batch --time --only --processes --kilobytes > iotop-`date +%Y%m%dT%H%M%S`.log | sudo free -m -s 1 -t -l > free-`date +%Y%m%dT%H%M%S`.log | sudo dstat -gmrsty -c -C 0,1,2,3,total --output dstat-`date +%Y%m%dT%H%M%S`.log | sudo /home/lefv2603/projects/jetson-inference/build/aarch64/bin/segnet-camera --network=fcn-resnet18-deepscene > segnet-camera.output-`date +%Y%m%dT%H%M%S`.log

Traitement

NVidia Tegrastats

tegrastats peut produire différentes traces, selon les éléments hardware qui sont mis à contribution.

trace 40

la trace produite par NVidia

1594343173 RAM 2410/3956MB (lfb 108x4MB) SWAP 314/1978MB (cached 7MB) CPU [1%@1479,0%@1479,1%@1479,1%@1479] EMC_FREQ 0% GR3D_FREQ 0% PLL@27C CPU@29.5C PMIC@100C GPU@29C AO@38.5C thermal@29C POM_5V_IN 2159/2159 POM_5V_GPU 166/166 POM_5V_CPU 332/332

quelque nettoyage, et arrangement en colonne, pour une meilleure visibilité

$ cat tegrastats-20200709T230009.log | sed -E 's/[()%\[]|MB|]|kB/ /g' | sed -E 's/[,@]|C / /g' | sed -E 's/ /\t/g' | sed -E 's/\t+/\t/g'
# dans Google sheet, copy => paste & transpose
1594343173
RAM MB
2410/3956
lfb Largest Free Block 
108x4
SWAP MB
314/1978
cached MB
7
CPU % Mhz
1
1479
0
1479
1
1479
1
1479
EMC_FREQ %/Mhz
0
GR3D_FREQ %/Mhz
0
PLL Celcius
27
CPU Celcius
29.5
PMIC Celcius
100
GPU Celcius
29
AO Celcius
38.5
thermal Celcius
29
POM_5V_IN milliwatts cur/avg
2159/2159
POM_5V_GPU milliwatts cur/avg
166/166 
POM_5V_CPU milliwatts cur/avg
332/332

et le rendu final, avec uniquement les valeurs:

$ cat tegrastats-20200709T230009.log | sed -E 's/[()%\[]|MB|]|kB/ /g' | sed -E 's/[,@]|C / /g' | sed -E 's/ /\t/g' | sed -E 's/\t+/\t/g' | awk -F$"\t" \
'BEGIN { OFS="," }; \
{ \
	split($3,RAM,"/"); \
	split($7,SWAP,"/"); \
	split($36,POM_5V_IN,"/"); \
	split($38,POM_5V_GPU,"/"); \
	split($40,POM_5V_CPU,"/"); \
	print $1,RAM[1],RAM[2],SWAP[1],SWAP[2],$9,$11,$12,$13,$14,$15,$16,$17,$18,$20,$22,$24,$26,$28,$30,$32,$34,POM_5V_IN[1],POM_5V_IN[2],POM_5V_GPU[1],POM_5V_GPU[2],POM_5V_CPU[1],POM_5V_CPU[2]; \
};
END {}; \'

retourne le résultat suivant:

1594343173,2410,3956,314,1978,7,1,1479,0,1479,1,1479,1,1479,0,0,27,29.5,100,29,38.5,29,2159,2159,166,166,332,332

header:

unixepochtime,ramused,ramtotal,swapused,swaptotal,cached,cpu1per,cpu1freq,cpu2per,cpu2freq,cpu3per,cpu3freq,cpu4per,cpu4freq,emcfreq,gr3dfreq,plltemp,cputemp,pmictemp,gputemp,aotemp,thermal,pom5vincurr,pom5vinavg,pom5vgpucurr,pom5vgpuavg,pom5vcpucurr,pom5vcpuavg

commande complète, avec sauvegarde dans un fichier csv

FILENAME="tegrastats-20200719T220319.log";echo "unixepochtime,ramused,ramtotal,swapused,swaptotal,cached,cpu1per,cpu1freq,cpu2per,cpu2freq,cpu3per,cpu3freq,cpu4per,cpu4freq,emcfreq,gr3dfreq,plltemp,cputemp,pmictemp,gputemp,aotemp,thermal,pom5vincurr,pom5vinavg,pom5vgpucurr,pom5vgpuavg,pom5vcpucurr,pom5vcpuavg" > ${FILENAME}.csv; cat ${FILENAME} | sed -E 's/[()%\[]|MB|]|kB/ /g' | sed -E 's/[,@]|C / /g' | sed -E 's/ /\t/g' | sed -E 's/\t+/\t/g' | awk -F$"\t" 'BEGIN { OFS="," }; \
{ \
split($3,RAM,"/"); \
split($7,SWAP,"/"); \
split($36,POM_5V_IN,"/"); \
split($38,POM_5V_GPU,"/"); \
split($40,POM_5V_CPU,"/"); \
print $1,RAM[1],RAM[2],SWAP[1],SWAP[2],$9,$11,$12,$13,$14,$15,$16,$17,$18,$20,$22,$24,$26,$28,$30,$32,$34,POM_5V_IN[1],POM_5V_IN[2],POM_5V_GPU[1],POM_5V_GPU[2],POM_5V_CPU[1],POM_5V_CPU[2]; \
};
END {}; \' >> ${FILENAME}.csv

Résultat

unixepochtime,ramused,ramtotal,swapused,swaptotal,cached,cpu1per,cpu1freq,cpu2per,cpu2freq,cpu3per,cpu3freq,cpu4per,cpu4freq,emcfreq,gr3dfreq,plltemp,cputemp,pmictemp,gputemp,aotemp,thermal,pom5vincurr,pom5vinavg,pom5vgpucurr,pom5vgpuavg,pom5vcpucurr,pom5vcpuavg
1595210599,1981,3956,0,1978,0,1,1479,1,1479,1,1479,0,1479,0,0,29,30,100,30,39.5,30.25,2201,2201,166,166,332,332

trace 48

Autre version avec la segmentation avec la caméra, oui il y a les stats suivantes en plus: IRAM, IRAM-lfb, EMC_FREQ Mhz, GR3D_FREQ Mhz, APE

la trace produite par NVidia

1594350022 RAM 2745/3956MB (lfb 11x4MB) SWAP 342/1978MB (cached 14MB) IRAM 0/252kB(lfb 252kB) CPU [6%@1479,0%@1479,1%@1479,3%@1479] EMC_FREQ 2%@1600 GR3D_FREQ 0%@921 APE 25 PLL@28.5C CPU@29.5C PMIC@100C GPU@30C AO@38.5C thermal@29.5C POM_5V_IN 2322/2343 POM_5V_GPU 165/182 POM_5V_CPU 373/400

quelque nettoyage, et arrangement en colonne, pour une meilleure visibilité

$ cat tegrastats-20200709T230009.log | sed -E 's/[()%\[]|MB|]|kB/ /g' | sed -E 's/[,@]|C / /g' | sed -E 's/ /\t/g' | sed -E 's/\t+/\t/g'
# dans Google sheet, copy => paste & transpose
1594350022
RAM
2745/3956
lfb
11x4
SWAP
342/1978
cached
14
IRAM
0/252
lfb
252
CPU
6
1479
0
1479
1
1479
3
1479
EMC_FREQ
2
1600
GR3D_FREQ
0
921
APE
25
PLL
28.5
CPU
29.5
PMIC
100
GPU
30
AO
38.5
thermal
29.5
POM_5V_IN
2322/2343
POM_5V_GPU
165/182
POM_5V_CPU
373/400

et le rendu final, avec uniquement les valeurs:

$ cat tegrastats-20200709T230009.log | sed -E 's/[()%\[]|MB|]|kB/ /g' | sed -E 's/[,@]|C / /g' | sed -E 's/ /\t/g' | sed -E 's/\t+/\t/g' | awk -F$"\t" \
'BEGIN { OFS="," }; \
{ \
  split($3,RAM,"/"); \
  split($7,SWAP,"/"); \
  split($11,IRAM,"/"); \
  split($44,POM_5V_IN,"/"); \
  split($46,POM_5V_GPU,"/"); \
  split($48,POM_5V_CPU,"/"); \
  print $1,RAM[1],RAM[2],SWAP[1],SWAP[2],$9,IRAM[1],IRAM[2],$15,$16,$17,$18,$19,$20,$21,$22,$24,$25,$27,$28,$30,$32,$34,$36,$38,$40,$42,POM_5V_IN[1],POM_5V_IN[2],POM_5V_GPU[1],POM_5V_GPU[2],POM_5V_CPU[1],POM_5V_CPU[2]; \
};
END {}; \'

donne le résultat suivant:

1594350022,2745,3956,342,1978,14,0,252,6,1479,0,1479,1,1479,3,1479,2,1600,0,921,25,28.5,29.5,100,30,38.5,29.5,2322,2343,165,182,373,400

header:

unixepochtime,ramused,ramtotal,swapused,swaptotal,cached,iramused,iramtotal,cpu1per,cpu1freq,cpu2per,cpu2freq,cpu3per,cpu3freq,cpu4per,cpu4freq,emcper,emcfreq,gr3dper,gr3dfreq,ape,plltemp,cputemp,pmictemp,gputemp,aotemp,thermal,pom5vincurr,pom5vinavg,pom5vgpucurr,pom5vgpuavg,pom5vcpucurr,pom5vcpuavg

commande complète, avec sauvegarde dans un fichier csv

FILENAME="tegrastats-20200719T220319.log";echo "unixepochtime,ramused,ramtotal,swapused,swaptotal,cached,iramused,iramtotal,cpu1per,cpu1freq,cpu2per,cpu2freq,cpu3per,cpu3freq,cpu4per,cpu4freq,emcper,emcfreq,gr3dper,gr3dfreq,ape,plltemp,cputemp,pmictemp,gputemp,aotemp,thermal,pom5vincurr,pom5vinavg,pom5vgpucurr,pom5vgpuavg,pom5vcpucurr,pom5vcpuavg" > "${FILENAME}.48.csv"
      cat "${FILENAME}" | sed -E 's/[()%\[]|MB|]|kB/ /g' | sed -E 's/[,@]|C / /g' | sed -E 's/ /\t/g' | sed -E 's/\t+/\t/g' | awk -F$"\t" \
      'BEGIN { OFS="," }; \
      { \
          split($3,RAM,"/"); \
          split($7,SWAP,"/"); \
          split($11,IRAM,"/"); \
          split($44,POM_5V_IN,"/"); \
          split($46,POM_5V_GPU,"/"); \
          split($48,POM_5V_CPU,"/"); \
          print $1,RAM[1],RAM[2],SWAP[1],SWAP[2],$9,IRAM[1],IRAM[2],$15,$16,$17,$18,$19,$20,$21,$22,$24,$25,$27,$28,$30,$32,$34,$36,$38,$40,$42,POM_5V_IN[1],POM_5V_IN[2],POM_5V_GPU[1],POM_5V_GPU[2],POM_5V_CPU[1],POM_5V_CPU[2]; \
      };
      END {}; \' >> "${FILENAME}.48.csv"

résultat:

unixepochtime,ramused,ramtotal,swapused,swaptotal,cached,iramused,iramtotal,cpu1per,cpu1freq,cpu2per,cpu2freq,cpu3per,cpu3freq,cpu4per,cpu4freq,emcper,emcfreq,gr3dper,gr3dfreq,ape,plltemp,cputemp,pmictemp,gputemp,aotemp,thermal,pom5vincurr,pom5vinavg,pom5vgpucurr,pom5vgpuavg,pom5vcpucurr,pom5vcpuavg
1595289757,2230,3956,0,1978,0,0,252,4,1479,5,1479,50,1479,7,1479,3,1600,0,921,25,28.5,30.5,100,30,39.5,30.75,2488,2488,165,165,579,579
1595289758,2229,3956,0,1978,0,0,252,3,1479,4,1479,18,1479,3,1479,3,1600,0,921,25,28.5,30,100,30,39,30,2525,2506,165,165,621,600
1595289759,2229,3956,0,1978,0,0,252,3,1479,2,1479,17,1479,3,1479,3,1600,0,921,25,29,30.5,100,30,39,30,2608,2540,165,165,703,634

free -m

cette commande

$ cat free.log | sed -Ee 's/[^[:digit:]]+/ /g' -Ee 's/^ //g' -Ee '/^$/d' | sed -e 'N;N;N;N; s/\n/ /g' -Ee 's/ /,/g'

transforme ceci :

total        used        free      shared  buff/cache   available
1594438559 Mem:           3956        2377         535         124        1044        1572
Low:           3956        3421         535
High:             0           0           0
Swap:          1978         388        1590
Total:         5934        2765        2125

total        used        free      shared  buff/cache   available
1594438560 Mem:           3956        2377         534         124        1044        1571
Low:           3956        3422         534
High:             0           0           0
Swap:          1978         388        1590
Total:         5934        2765        2124

total        used        free      shared  buff/cache   available
1594438561 Mem:           3956        2380         531         124        1044        1568
Low:           3956        3424         531
High:             0           0           0
Swap:          1978         388        1590
Total:         5934        2768        2121

total        used        free      shared  buff/cache   available
1594438562 Mem:           3956        2376         535         124        1044        1572
Low:           3956        3420         535
High:             0           0           0
Swap:          1978         388        1590
Total:         5934        2764        2125

en cela

1594438559,3956,2377,535,124,1044,1572,3956,3421,535,0,0,0,1978,388,1590,5934,2765,2125
1594438560,3956,2377,534,124,1044,1571,3956,3422,534,0,0,0,1978,388,1590,5934,2765,2124
1594438561,3956,2380,531,124,1044,1568,3956,3424,531,0,0,0,1978,388,1590,5934,2768,2121
1594438562,3956,2376,535,124,1044,1572,3956,3420,535,0,0,0,1978,388,1590,5934,2764,2125

header:

unixepochtime,memtotal,memused,memfree,memshared,membufcache,memavailable,lowtotal,lowused,lowfree,hightotal,highused,highfree,swaptotal,swapused,swapfree,totaltotal,totalused,totalfree

commande complète, avec sauvegarde dans un fichier

FILENAME="free-20200719T231632.log"; echo "unixepochtime,memtotal,memused,memfree,memshared,membufcache,memavailable,lowtotal,lowused,lowfree,hightotal,highused,highfree,swaptotal,swapused,swapfree,totaltotal,totalused,totalfree" > ${FILENAME}.csv; cat ${FILENAME} | sed -Ee 's/[^[:digit:]]+/ /g' -Ee 's/^ //g' -Ee '/^$/d' | sed -e 'N;N;N;N; s/\n/ /g' -Ee 's/ /,/g' >> ${FILENAME}.csv

résultat

unixepochtime,memtotal,memused,memfree,memshared,membufcache,memavailable,lowtotal,lowused,lowfree,hightotal,highused,highfree,swaptotal,swapused,swapfree,totaltotal,totalused,totalfree
1595214996,3956,2807,354,246,793,770,3956,3601,354,0,0,0,1978,40,1937,5934,2848,2291
1595214997,3956,2805,356,246,793,769,3956,3599,356,0,0,0,1978,40,1937,5934,2846,2293
1595214998,3956,2805,356,246,793,769,3956,3599,356,0,0,0,1978,40,1937,5934,2846,2293
1595214999,3956,2805,356,246,793,770,3956,3599,356,0,0,0,1978,40,1937,5934,2846,2293

iotop

segnet-camera

cette commande :

$ grep -E "segnet" iotop.log | sed -E 's/%|(K\/s)//g' | awk -F" " -v logdate="Jul 9, 2020" \
'BEGIN { OFS="," }; \
{ \
	split($1,time,":"); \
	cmd ="date \"+%s\" -d \""logdate" "time[1]":"time[2]":"time[3]"\""; \
	cmd | getline var; \
	print var,$5,$6,$7,$8; \
};
END {}; \'

transforme ceci:

30:34 11592 ?sys root     5697.11 K/s    0.00 K/s  0.00 %  0.88 % segnet-camera --network=fcn-resnet18-deepscene
30:35 11592 ?sys root     31904.62 K/s    0.00 K/s  0.00 %  8.71 % segnet-camera --network=fcn-resnet18-deepscene
30:36 11592 ?sys root     2969.61 K/s    0.00 K/s  0.00 %  6.83 % segnet-camera --network=fcn-resnet18-deepscene
30:38 11592 ?sys root     1225.92 K/s    6.02 K/s  0.00 %  1.49 % segnet-camera --network=fcn-resnet18-deepscene
30:39 11592 ?sys root      824.32 K/s    6.81 K/s  0.00 %  0.49 % segnet-camera --network=fcn-resnet18-deepscene
30:40 11592 ?sys root        0.00 K/s    3.40 K/s  0.00 %  0.00 % segnet-camera --network=fcn-resnet18-deepscene
30:41 11592 ?sys root        0.00 K/s    6.69 K/s  0.00 %  0.00 % segnet-camera --network=fcn-resnet18-deepscene
30:42 11592 ?sys root        0.00 K/s    3.42 K/s  0.00 %  0.00 % segnet-camera --network=fcn-resnet18-deepscene
30:43 11592 ?sys root        0.00 K/s    6.85 K/s  0.00 %  0.00 % segnet-camera --network=fcn-resnet18-deepscene

en cela:

1594341034,5697.11,0.00,0.00,0.88
1594341035,31904.62,0.00,0.00,8.71
1594341036,2969.61,0.00,0.00,6.83
1594341038,1225.92,6.02,0.00,1.49
1594341039,824.32,6.81,0.00,0.49
1594341040,0.00,3.40,0.00,0.00
1594341041,0.00,6.69,0.00,0.00
1594341042,0.00,3.42,0.00,0.00
1594341043,0.00,6.85,0.00,0.00

header:

unixepochtime,diskread,diskwrite,swapin,io

commande complète, avec sauvegarde dans un fichier

$ FILENAME="iotop-20200719T224932.log"; echo "unixepochtime,diskread,diskwrite,swapin,io" > ${FILENAME}.segnet-camera.csv; grep -E "segnet" iotop.log | sed -E 's/%|(K\/s)//g' | awk -F" " -v logdate="Jul 9, 2020" \
'BEGIN { OFS="," }; \
{ \
	split($1,time,":"); \
	cmd ="date \"+%s\" -d \""logdate" "time[1]":"time[2]":"time[3]"\""; \
	cmd | getline var; \
	print var,$5,$6,$7,$8; \
};
END {}; \' >> ${FILENAME}.segnet-camera.csv

résultat

unixepochtime,diskread,diskwrite,swapin,io
1594341034,5697.11,0.00,0.00,0.88
1594341035,31904.62,0.00,0.00,8.71
1594341036,2969.61,0.00,0.00,6.83
1594341038,1225.92,6.02,0.00,1.49
1594341039,824.32,6.81,0.00,0.49
1594341040,0.00,3.40,0.00,0.00
1594341041,0.00,6.69,0.00,0.00
1594341042,0.00,3.42,0.00,0.00
1594341043,0.00,6.85,0.00,0.00

Total disk

cette commande :

$ FILENAME="iotop-20200719T224932.log"; grep -E "Total DISK" $FILENAME | sed -E 's/\||(K\/s)|(Total DISK READ :)|(Total DISK WRITE :)//g' | awk -F" " -v logdate="Jul 9, 2020" 'BEGIN { OFS="," }; \
{ \                  
split($1,time,":"); \                                                
cmd ="date \"+%s\" -d \""logdate" "time[1]":"time[2]":"time[3]"\""; \
cmd | getline var; \
print var,$2,$3; \
};        
END {}; \'

transforme ceci:

30:34 Total DISK READ :   13141.87 K/s | Total DISK WRITE :     116.97 K/s
30:35 Total DISK READ :   35516.96 K/s | Total DISK WRITE :       2.93 K/s
30:36 Total DISK READ :   20181.52 K/s | Total DISK WRITE :      11.82 K/s
30:38 Total DISK READ :   22590.62 K/s | Total DISK WRITE :      24.10 K/s
30:39 Total DISK READ :   19208.00 K/s | Total DISK WRITE :      10.22 K/s
30:40 Total DISK READ :   17365.75 K/s | Total DISK WRITE :       6.80 K/s
30:41 Total DISK READ :   24771.96 K/s | Total DISK WRITE :      10.03 K/s
30:42 Total DISK READ :      10.25 K/s | Total DISK WRITE :      54.67 K/s
30:43 Total DISK READ :    5607.93 K/s | Total DISK WRITE :       6.85 K/s

en cela:

1594341034,13141.87,116.97
1594341035,35516.96,2.93
1594341036,20181.52,11.82
1594341038,22590.62,24.10
1594341039,19208.00,10.22
1594341040,17365.75,6.80
1594341041,24771.96,10.03
1594341042,10.25,54.67
1594341043,5607.93,6.85

header:

unixepochtime,totaldiskread,totaldiskwrwite

commande complète, avec sauvegarde dans un fichier csv

$ FILENAME="iotop-20200719T224932.log"; echo "unixepochtime,totaldiskread,totaldiskwrwite" > ${FILENAME}.totaldisk.csv; grep -E "Total DISK" $FILENAME | sed -E 's/\||(K\/s)|(Total DISK READ :)|(Total DISK WRITE :)//g' | awk -F" " -v logdate="Jul 9, 2020" 'BEGIN { OFS="," }; \
{ \                  
split($1,time,":"); \                                                
cmd ="date \"+%s\" -d \""logdate" "time[1]":"time[2]":"time[3]"\""; \
cmd | getline var; \
print var,$2,$3; \
};        
END {}; \' >> ${FILENAME}.totaldisk.csv

résultat

unixepochtime,totaldiskread,totaldiskwrwite
1594349372,0.00,0.00
1594349374,0.00,10.57
1594349375,0.00,0.00
1594349376,0.00,0.00
1594349377,0.00,0.00

Collecteur de stats

Script qui collecte les trois métriques: tegrastats, free -m et iotop.

Le paramètre est le chemin ou sont enregistrés les fichiers de stats, et sert aussi comme nom du test.

#!/bin/bash

if [ $# -eq 0 ]
  then
    echo "No arguments supplied"
    exit
fi

if [ -z "$1" ]
  then
  echo "No argument supplied"
fi

REASON=$1
[ -d ${REASON} ] && rm -rf ${REASON}

mkdir ${REASON}

# tegrastats stat collect

TEGRASTATS_FILENAME=${REASON}/"tegrastats-`date +%Y%m%dT%H%M%S`.log"
sudo tegrastats | while read a; do echo "$a" | awk -v now="$(date +%s)" '{OFS=" "}{print now" "$0}' >> ${TEGRASTATS_FILENAME};  done &

# free -m stat collect
FREE_FILENAME=${REASON}/"free-`date +%Y%m%dT%H%M%S`.log"
sudo free -m -s 1 -t -l | while read a; do echo "$a" | awk -v now="$(date +%s)" '{OFS=" "}{if(NF==7) {print now" "$0;} else {print $0;}}' >> ${FREE_FILENAME};  done &

# iotop stat collect
IOTOP_FILENAME=${REASON}/"iotop-`date +%Y%m%dT%H%M%S`.log"
sudo iotop --batch --time --only --processes --kilobytes > ${IOTOP_FILENAME}

wait
echo "done"

Transformateur en CSV

Script qui prend en entré le chemin où sont enregistrés les fichiers de stats, et qui sert aussi comme nom du test.

Transforme les fichiers au format CSV.

#!/bin/bash

if [ $# -eq 0 ]
  then
    echo "No arguments supplied"
    exit
fi

if [ -z "$1" ]
  then
  echo "No argument supplied"
fi

REASON=$1

# tegrastats stat collect
FILENAME="`ls -1 ${REASON}/tegrastats-*.log`" 
# test if IRAM & APE stats present (48s stats)
if [ $(grep -cE "IRAM.*APE" ${FILENAME}) > 0 ]
then
        echo "unixepochtime,ramused,ramtotal,swapused,swaptotal,cached,iramused,iramtotal,cpu1per,cpu1freq,cpu2per,cpu2freq,cpu3per,cpu3freq,cpu4per,cpu4freq,emcper,emcfreq,gr3dper,gr3dfreq,ape,plltemp,cputemp,pmictemp,gputemp,aotemp,thermal,pom5vincurr,pom5vinavg,pom5vgpucurr,pom5vgpuavg,pom5vcpucurr,pom5vcpuavg" > "${FILENAME}.48.csv"
        cat "${FILENAME}" | sed -E 's/[()%\[]|MB|]|kB/ /g' | sed -E 's/[,@]|C / /g' | sed -E 's/ /\t/g' | sed -E 's/\t+/\t/g' | awk -F$"\t" \
        'BEGIN { OFS="," }; \
        { \
	        split($3,RAM,"/"); \
	        split($7,SWAP,"/"); \
	        split($11,IRAM,"/"); \
	        split($44,POM_5V_IN,"/"); \
	        split($46,POM_5V_GPU,"/"); \
	        split($48,POM_5V_CPU,"/"); \
	        print $1,RAM[1],RAM[2],SWAP[1],SWAP[2],$9,IRAM[1],IRAM[2],$15,$16,$17,$18,$19,$20,$21,$22,$24,$25,$27,$28,$30,$32,$34,$36,$38,$40,$42,POM_5V_IN[1],POM_5V_IN[2],POM_5V_GPU[1],POM_5V_GPU[2],POM_5V_CPU[1],POM_5V_CPU[2]; \
        };
        END {}; \' >> "${FILENAME}.48.csv"
else
        echo "unixepochtime,ramused,ramtotal,swapused,swaptotal,cached,cpu1per,cpu1freq,cpu2per,cpu2freq,cpu3per,cpu3freq,cpu4per,cpu4freq,emcfreq,gr3dfreq,plltemp,cputemp,pmictemp,gputemp,aotemp,thermal,pom5vincurr,pom5vinavg,pom5vgpucurr,pom5vgpuavg,pom5vcpucurr,pom5vcpuavg" > "${FILENAME}.40.csv"
        cat "${FILENAME}" | sed -E 's/[()%\[]|MB|]|kB/ /g' | sed -E 's/[,@]|C / /g' | sed -E 's/ /\t/g' | sed -E 's/\t+/\t/g' | awk -F$"\t" 'BEGIN { OFS="," }; \
        { \
        split($3,RAM,"/"); \
        split($7,SWAP,"/"); \
        split($36,POM_5V_IN,"/"); \
        split($38,POM_5V_GPU,"/"); \
        split($40,POM_5V_CPU,"/"); \
        print $1,RAM[1],RAM[2],SWAP[1],SWAP[2],$9,$11,$12,$13,$14,$15,$16,$17,$18,$20,$22,$24,$26,$28,$30,$32,$34,POM_5V_IN[1],POM_5V_IN[2],POM_5V_GPU[1],POM_5V_GPU[2],POM_5V_CPU[1],POM_5V_CPU[2]; \
        };
        END {}; \' >> "${FILENAME}.40.csv"
fi

# free -m stat collect
FILENAME="`ls -1 ${REASON}/free-*.log`" 
echo "unixepochtime,memtotal,memused,memfree,memshared,membufcache,memavailable,lowtotal,lowused,lowfree,hightotal,highused,highfree,swaptotal,swapused,swapfree,totaltotal,totalused,totalfree" > "${FILENAME}.csv"
cat "${FILENAME}" | sed -Ee 's/[^[:digit:]]+/ /g' -Ee 's/^ //g' -Ee '/^$/d' | sed -e 'N;N;N;N; s/\n/ /g' -Ee 's/ /,/g' >> "${FILENAME}.csv"

# iotop
FILENAME="`ls -1 ${REASON}/iotop-*.log`"
# iotop segnet-camera stat to CSV transformation
echo "unixepochtime,diskread,diskwrite,swapin,io" > "${FILENAME}.segnet-camera.csv"
grep -E "segnet-camera" ${FILENAME} | sed -E 's/%|(K\/s)//g' | awk -F" " -v logdate="Jul 20, 2020" \
'BEGIN { OFS="," }; \
{ \
	split($1,time,":"); \
	cmd ="date \"+%s\" -d \""logdate" "time[1]":"time[2]":"time[3]"\""; \
	cmd | getline var; \
	print var,$5,$6,$7,$8; \
};
END {}; \' >> "${FILENAME}.segnet-camera.csv"

# iotop total disk stat to CSV transformation
echo "unixepochtime,totaldiskread,totaldiskwrwite" > "${FILENAME}.totaldisk.csv"
grep -E "Total DISK" "$FILENAME" | sed -E 's/\||(K\/s)|(Total DISK READ :)|(Total DISK WRITE :)//g' | awk -F" " -v logdate="Jul 9, 2020" 'BEGIN { OFS="," }; \
{ \                  
split($1,time,":"); \                                                
cmd ="date \"+%s\" -d \""logdate" "time[1]":"time[2]":"time[3]"\""; \
cmd | getline var; \
print var,$2,$3; \
};        
END {}; \' >> "${FILENAME}.totaldisk.csv"

echo "done"

Diagramme

# -*- coding: utf-8 -*-
"""Untitled1.ipynb

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/1oqDbgStUGPkgC3qx3uzMLaNze_MmmTig
"""

# importing the required module 
import matplotlib.pyplot as plt 
  
# x axis values 
x = [1,2,3] 
# corresponding y axis values 
y = [2,4,1] 
  
# plotting the points  
plt.plot(x, y) 
  
# naming the x axis 
plt.xlabel('x - axis') 
# naming the y axis 
plt.ylabel('y - axis') 
  
# giving a title to my graph 
plt.title('My first graph!') 
  
# function to show the plot 
plt.show()



import pandas as pd
import matplotlib.pyplot as plt

plt.style.use('ggplot')

df = pd.read_csv('/content/free-20200721T215303.log.csv')
df.head()
#df.plot(x='unixepochtime', y=['memused'])

fig, ax = plt.subplots(figsize=(10,5))

zeroorigints = [ts - timestamps[0] for ts in df['unixepochtime']]
ax.plot(zeroorigints, df['totaltotal'], label='Memoire totale Mbytes')
ax.plot(zeroorigints, df['totalused'], label='Memoire utilise Mbytes')
ax.plot(zeroorigints, df['totalfree'], label='Memoire libre Mbytes')

ax.axvline(x=0, color='k')
ax.axhline(y=0, color='k')

plt.legend(loc='lower right', bbox_to_anchor=(1, 0.1))
plt.show()

import pandas as pd
import numpy as np

plt.style.use('ggplot')

data = np.loadtxt('/content/free-20200721T215303.log.csv', delimiter=',', skiprows=1)

fig, ax = plt.subplots(figsize=(10,5))

unixepochtime = data[:,0]
totaltotal = data[:,-3]
totalused = data[:,-2]
totalfree = data[:,-1]
zeroorigints = [ts - timestamps[0] for ts in unixepochtime]
ax.plot(zeroorigints, totaltotal, label='Memoire totale Mbytes')
ax.plot(zeroorigints, totalused, label='Memoire utilise Mbytes')
ax.plot(zeroorigints, totalfree, label='Memoire libre Mbytes')

plt.xlabel("Periode en seconde")
plt.ylabel("Utilisation en MBytes")
ax.axvline(x=0, color='k')
ax.axhline(y=0, color='k')
plt.legend(loc='lower right', bbox_to_anchor=(1, 0.1))
plt.show()

import pandas as pd
import matplotlib.pyplot as plt

plt.style.use('ggplot')

df = pd.read_csv('/content/tegrastats-20200721T215303.log.48.csv')
df.head()

fig, ax = plt.subplots(figsize=(10,5))

zeroorigints = [ts - timestamps[0] for ts in df['unixepochtime']]
ax.plot(zeroorigints, df['plltemp'], label='PLL C')
ax.plot(zeroorigints, df['cputemp'], label='CPU C')
#ax.plot(zeroorigints, df['pmictemp'], label='PMIC C')
ax.plot(zeroorigints, df['gputemp'], label='GPU C')
ax.plot(zeroorigints, df['aotemp'], label='AO C')
ax.plot(zeroorigints, df['thermal'], label='Thermal C')

plt.xlabel("Periode en seconde")
plt.ylabel("Temperatude en degre Celcius")
plt.ylim(top=45)
ax.axhline(y=0, color='k')
ax.axvline(x=0, color='k')
plt.legend(loc='lower right', bbox_to_anchor=(1, 0.1))
plt.show()

fig, ax = plt.subplots(figsize=(10,5))

zeroorigints = [ts - timestamps[0] for ts in df['unixepochtime']]
ax.plot(zeroorigints, df['cpu1per'], label='CPU 1 %')
ax.plot(zeroorigints, df['cpu2per'], label='CPU 2 %')
ax.plot(zeroorigints, df['cpu3per'], label='CPU 3 %')
ax.plot(zeroorigints, df['cpu4per'], label='CPU 4 %')
ax.plot(zeroorigints, df['emcper'], label='EMC %')

plt.xlabel("Periode en seconde")
plt.ylabel("Frequence en %")
plt.ylim(top=100)
ax.axhline(y=0, color='k')
ax.axvline(x=0, color='k')
plt.legend(loc='lower right', bbox_to_anchor=(1, 0.1))
plt.show()

fig, ax = plt.subplots(figsize=(10,5))

zeroorigints = [ts - timestamps[0] for ts in df['unixepochtime']]
ax.plot(zeroorigints, df['pom5vincurr'], label='5V IN')
ax.plot(zeroorigints, df['pom5vinavg'], label='5V IN moy.')
ax.plot(zeroorigints, df['pom5vgpucurr'], label='5V GPU')
ax.plot(zeroorigints, df['pom5vgpuavg'], label='5V GPU moy.')
ax.plot(zeroorigints, df['pom5vcpucurr'], label='5V CPU')
ax.plot(zeroorigints, df['pom5vcpuavg'], label='5V CPU moy.')

plt.xlabel("Periode en seconde")
plt.ylabel("Consommation en millivolt")
#plt.ylim(top=100)
ax.axhline(y=0, color='k')
ax.axvline(x=0, color='k')
plt.legend(loc='lower right', bbox_to_anchor=(1, 0.1))
plt.show()

import pandas as pd
import matplotlib.pyplot as plt

plt.style.use('ggplot')

df = pd.read_csv('/content/iotop-20200721T215303.log.totaldisk.csv')
df.head()

fig, ax = plt.subplots(figsize=(10,5))

zeroorigints = [ts - timestamps[0] for ts in df['unixepochtime']]
ax.plot(zeroorigints, df['totaldiskread'], label='Total disk read Kbytes')
ax.plot(zeroorigints, df['totaldiskwrwite'], label='Total disk write Kbytes')

plt.xlabel("Periode en seconde")
plt.ylabel("bandwidth between processes and kernel threads in kilobytes")
#plt.ylim(top=100)
#ax.axhline(y=0, color='k')
#ax.axvline(x=0, color='k')
plt.legend(loc='lower right', bbox_to_anchor=(1, 0.1))
plt.show()