nicolargo/glances
1
1
Fork 0
mirror of https://github.com/nicolargo/glances.git synced 2024-12-25 10:12:55 +03:00
Code Issues Projects Releases Wiki Activity

NVIDIA GPU temperature #1523

Browse Source
This commit is contained in:
nicolargo 2019-11-09 18:01:39 +01:00
parent e7cc51fab9
commit d434333dfe
5 changed files with 67 additions and 30 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
docs/aoa/gpu.rst
View File

@ -4,11 +4,15 @@ GPU
===
.. note::
You need to install the `nvidia-ml-py3`_ library on your system.
You need to install the `py3nvml`_ library on your system.
Or `nvidia-ml-py3`_ for Glances 3.1.3 or lower.
The GPU stats are shown as a percentage of value and for the configured
refresh time. The total GPU usage is displayed on the first line, the
memory consumption on the second one.
refresh time. It displays:
- total GPU usage
- memory consumption
- temperature (Glances 3.1.4 or higher)
.. image:: ../_static/gpu.png
@ -45,4 +49,5 @@ GPU (PROC/MEM) Status
``>90%`` ``CRITICAL``
============== ============
.. _nvidia-ml-py3: https://pypi.python.org/pypi/nvidia-ml-py3
.. _py3nvml: https://pypi.org/project/py3nvml/
.. _nvidia-ml-py3: https://pypi.org/project/nvidia-ml-py3/

5
glances/compat.py
View File

@ -234,3 +234,8 @@ def time_serie_subsample(data, sampling):
t_subsampled = [t[s * sampling_length:(s + 1) * sampling_length][0] for s in range(0, sampling)]
v_subsampled = [mean(v[s * sampling_length:(s + 1) * sampling_length]) for s in range(0, sampling)]
return list(zip(t_subsampled, v_subsampled))
def to_fahrenheit(celsius):
"""Convert Celsius to Fahrenheit."""
return celsius * 1.8 + 32

70
glances/plugins/glances_gpu.py
View File

@ -19,12 +19,13 @@
"""GPU plugin (limited to NVIDIA chipsets)."""
from glances.compat import nativestr
from glances.compat import nativestr, to_fahrenheit
from glances.logger import logger
from glances.plugins.glances_plugin import GlancesPlugin
# In Glances 3.1.4 or higher, we use the py3nvml lib (see issue #1523)
try:
import pynvml
import py3nvml.py3nvml as pynvml
except Exception as e:
import_error_tag = True
# Display debu message if import KeyError
@ -86,17 +87,6 @@ class Plugin(GlancesPlugin):
# Init new stats
stats = self.get_init_value()
# !!! JUST FOR TEST (because i did not have any NVidia GPU... :()
# self.stats = [{"key": "gpu_id", "mem": None, "proc": 60, "gpu_id": 0, "name": "GeForce GTX 560 Ti"}]
# self.stats = [{"key": "gpu_id", "mem": 10, "proc": 60, "gpu_id": 0, "name": "GeForce GTX 560 Ti"}]
# self.stats = [{"key": "gpu_id", "mem": 48.64645, "proc": 60.73, "gpu_id": 0, "name": "GeForce GTX 560 Ti"},
# {"key": "gpu_id", "mem": 70.743, "proc": 80.28, "gpu_id": 1, "name": "GeForce GTX 560 Ti"},
# {"key": "gpu_id", "mem": 0, "proc": 0, "gpu_id": 2, "name": "GeForce GTX 560 Ti"}]
# self.stats = [{"key": "gpu_id", "mem": 48.64645, "proc": 60.73, "gpu_id": 0, "name": "GeForce GTX 560 Ti"},
# {"key": "gpu_id", "mem": None, "proc": 80.28, "gpu_id": 1, "name": "GeForce GTX 560 Ti"},
# {"key": "gpu_id", "mem": 0, "proc": 0, "gpu_id": 2, "name": "ANOTHER GPU"}]
# !!! TO BE COMMENTED
if not self.nvml_ready:
return self.stats
@ -120,7 +110,9 @@ class Plugin(GlancesPlugin):
# Alert
for i in self.stats:
# Init the views for the current GPU
self.views[i[self.get_key()]] = {'proc': {}, 'mem': {}}
self.views[i[self.get_key()]] = {'proc': {},
'mem': {},
'temperature': {}}
# Processor alert
if 'proc' in i:
alert = self.get_alert(i['proc'], header='proc')
@ -129,6 +121,10 @@ class Plugin(GlancesPlugin):
if 'mem' in i:
alert = self.get_alert(i['mem'], header='mem')
self.views[i[self.get_key()]]['mem']['decoration'] = alert
# Temperature alert
if 'temperature' in i:
alert = self.get_alert(i['temperature'], header='temperature')
self.views[i[self.get_key()]]['temperature']['decoration'] = alert
return True
@ -197,12 +193,36 @@ class Plugin(GlancesPlugin):
mean_mem_msg, self.get_views(item=gpu_stats[self.get_key()],
key='mem',
option='decoration')))
# New line
ret.append(self.curse_new_line())
# GPU TEMPERATURE
try:
mean_temperature = sum(s['temperature'] for s in self.stats if s is not None) / len(self.stats)
except TypeError:
mean_temperature_msg = '{:>4}'.format('N/A')
else:
unit = 'C'
if args.fahrenheit:
value = to_fahrenheit(i['value'])
unit = 'F'
mean_temperature_msg = '{:>3.0f}{}'.format(mean_temperature,
unit)
if len(self.stats) > 1:
msg = '{:13}'.format('temp mean:')
else:
msg = '{:13}'.format('temperature:')
ret.append(self.curse_add_line(msg))
ret.append(self.curse_add_line(
mean_temperature_msg, self.get_views(item=gpu_stats[self.get_key()],
key='temperature',
option='decoration')))
else:
# Multi GPU
# Temperature is not displayed in this mode...
for gpu_stats in self.stats:
# New line
ret.append(self.curse_new_line())
# GPU ID + PROC + MEM
# GPU ID + PROC + MEM + TEMPERATURE
id_msg = '{}'.format(gpu_stats['gpu_id'])
try:
proc_msg = '{:>3.0f}%'.format(gpu_stats['proc'])
@ -212,7 +232,9 @@ class Plugin(GlancesPlugin):
mem_msg = '{:>3.0f}%'.format(gpu_stats['mem'])
except ValueError:
mem_msg = '{:>4}'.format('N/A')
msg = '{}: {} mem: {}'.format(id_msg, proc_msg, mem_msg)
msg = '{}: {} mem: {}'.format(id_msg,
proc_msg,
mem_msg)
ret.append(self.curse_add_line(msg))
return ret
@ -222,7 +244,7 @@ class Plugin(GlancesPlugin):
stats = []
for index, device_handle in enumerate(self.device_handles):
device_stats = {}
device_stats = dict()
# Dictionnary key is the GPU_ID
device_stats['key'] = self.get_key()
# GPU id (for multiple GPU, start at 0)
@ -233,6 +255,8 @@ class Plugin(GlancesPlugin):
device_stats['mem'] = get_mem(device_handle)
# Processor consumption in %
device_stats['proc'] = get_proc(device_handle)
# Processor temperature in °C
device_stats['temperature'] = get_temperature(device_handle)
stats.append(device_stats)
return stats
@ -268,8 +292,7 @@ def get_device_name(device_handle):
def get_mem(device_handle):
"""Get GPU device memory consumption in percent."""
try:
memory_info = pynvml.nvmlDeviceGetMemoryInfo(device_handle)
return memory_info.used * 100.0 / memory_info.total
return pynvml.nvmlDeviceGetUtilizationRates(device_handle).memory
except pynvml.NVMLError:
return None
@ -280,3 +303,12 @@ def get_proc(device_handle):
return pynvml.nvmlDeviceGetUtilizationRates(device_handle).gpu
except pynvml.NVMLError:
return None
def get_temperature(device_handle):
"""Get GPU device CPU consumption in percent."""
try:
return pynvml.nvmlDeviceGetTemperature(device_handle,
pynvml.NVML_TEMPERATURE_GPU)
except pynvml.NVMLError:
return None

7
glances/plugins/glances_sensors.py
View File

@ -23,7 +23,7 @@ import psutil
import warnings
from glances.logger import logger
from glances.compat import iteritems
from glances.compat import iteritems, to_fahrenheit
from glances.timer import Counter
from glances.plugins.sensors.glances_batpercent import Plugin as BatPercentPlugin
from glances.plugins.sensors.glances_hddtemp import Plugin as HddTempPlugin
@ -33,11 +33,6 @@ SENSOR_TEMP_UNIT = 'C'
SENSOR_FAN_UNIT = 'R'
def to_fahrenheit(celsius):
"""Convert Celsius to Fahrenheit."""
return celsius * 1.8 + 32
class Plugin(GlancesPlugin):
"""Glances sensors plugin.

2
optional-requirements.txt
View File

@ -7,7 +7,7 @@ elasticsearch
influxdb
kafka-python
netifaces
nvidia-ml-py3
py3nvml
paho-mqtt
pika
potsdb