1PMDANVIDIA(1) General Commands Manual PMDANVIDIA(1)
2
3
4
6 pmdanvidia - nvidia gpu metrics domain agent (PMDA)
7
9 $PCP_PMDAS_DIR/nvidia/pmdanvidia [-d domain] [-l logfile] [-t interval]
10
12 pmdanvidia is a Performance Metrics Domain Agent (PMDA) which extracts
13 performance metrics describing the metrics available on NVIDIA GPU
14 cards via the NVML library.
15
16 The nvidia PMDA exports metrics that measure gpu activity, memory uti‐
17 lization, fan speed, etc on NVIDIA Tesla and Quadro cards. Metrics are
18 unlikely to be available for consumer class cards.
19
20 A brief description of the pmdanvidia command line options follows:
21
22 -d It is absolutely crucial that the performance metrics domain num‐
23 ber specified here is unique and consistent. That is, domain
24 should be different for every PMDA on the one host, and the same
25 domain number should be used for the same PMDA on all hosts.
26
27 -l Location of the log file. By default, a log file named nvidia.log
28 is written in the current directory of pmcd(1) when pmdanvidia is
29 started, i.e. $PCP_LOG_DIR/pmcd. If the log file cannot be cre‐
30 ated or is not writable, output is written to the standard error
31 instead.
32
33 -t Enables and sets a sampling interval for automatic refreshing of
34 metric values. The functionality is disabled by default, however
35 this option allows a time interval to be specified on which all
36 values are sampled - this has the effect of constantly updating
37 the accumulating metrics, with the goal of assisting client tools
38 such as pcp-atop(1) and pmlogger(1) to observe sub-sample time
39 changes in GPU and process state. Typically these tools have
40 longer sampling intervals, and can thus 'miss' activity happening
41 during their sampling interval.
42
44 The nvidia PMDA is not installed and available by default. If you want
45 to undo the installation, do the following as root:
46
47 # cd $PCP_PMDAS_DIR/nvidia
48 # ./Remove
49
50 If you want to establish access to the names, help text and values for
51 the nvidia performance metrics once more, after removal, do the follow‐
52 ing as root:
53
54 # cd $PCP_PMDAS_DIR/nvidia
55 # ./Install
56
57 pmdanvidia is launched by pmcd(1) and should never be executed di‐
58 rectly. The Install and Remove scripts notify pmcd(1) when the agent
59 is installed or removed.
60
62 $PCP_PMCDCONF_PATH
63 command line options used to launch pmdanvidia
64 $PCP_PMDAS_DIR/nvidia/help
65 default help text file for the nvidia metrics
66 $PCP_PMDAS_DIR/nvidia/Install
67 installation script for the pmdanvidia agent
68 $PCP_PMDAS_DIR/nvidia/Remove
69 undo installation script for the pmdanvidia agent
70 $PCP_LOG_DIR/pmcd/nvidia.log
71 default log file for error messages and other information
72 from pmdanvidia
73
75 Environment variables with the prefix PCP_ are used to parameterize the
76 file and directory names used by PCP. On each installation, the file
77 /etc/pcp.conf contains the local values for these variables. The
78 $PCP_CONF variable may be used to specify an alternative configuration
79 file, as described in pcp.conf(5).
80
82 PCPIntro(1), pcp-atop(1), pmcd(1), pmlogger(1), pcp.conf(5) and
83 pcp.env(5).
84
85
86
87Performance Co-Pilot PCP PMDANVIDIA(1)