1just-man-pages/condor_gpu_disGceonveerrayl(1C)ommajnudsst-Mmaannu-aplages/condor_gpu_discovery(1)
2
3
4

Name

6       condor_gpu_discovery Output GPU-related ClassAd attributes
7

Synopsis

9       condor_gpu_discovery -help
10
11       condor_gpu_discovery [ <options> ]
12

Description

14       condor_gpu_discovery  runs  discovery  software to determine the host's
15       GPU capabilities, which are output as ClassAd attributes.
16
17       This tool is not available for MAC OS platforms.
18
19       With no command line options, the single ClassAd attribute   DetectedG‐
20       PUs  is  printed.  If  the value is 0, no GPUs were detected. If one or
21       more GPUS were detected, the value is a string, presented  as  a  comma
22       and  space separated list of the GPUs discovered, where each is given a
23       name further used as the prefix string in other attribute names.  Where
24       there  is  more  than  one  GPU of a particular type, the prefix string
25       includes an integer value numbering the device;  these  integer  values
26       monotonically increase from 0. For example, a discovery of two GPUs may
27       output
28
29       DetectedGPUs="CUDA0, CUDA1"
30
31       Further command line options use  "CUDA" either with or without one  of
32       the integer values 0 or 1 as the prefix string in attribute names.
33

Options

35       -help
36
37          Print usage information and exit.
38
39
40
41       -properties
42
43          In  addition  to  the  DetectedGPUs attribute, display standard CUDA
44          attributes. Each of these attribute names will have a prefix  string
45          at  the  beginning of its name. For a host with more than one of the
46          same GPU type, those attribute values that are the same  across  all
47          of  the  GPUs  will not have an integer value in the prefix string .
48          The attributes are  Capability  ,   DeviceName  ,   DriverVersion  ,
49          ECCEnabled  ,   GlobalMemoryMb , and  RuntimeVersion . The displayed
50          standard Open CL attributes are  DeviceName ,  ECCEnabled  ,   Open‐
51          CLVersion , and  GlobalMemoryMb .
52
53
54
55       -extra
56
57          Display  the  additional attributes of Each of these attribute names
58          will have a prefix string at the beginning of its name.  ClockMhz  ,
59          ComputeUnits  , and  CoresPerCU for a CUDA device, and  ClockMhz and
60          ComputeUnits for an OCL device.
61
62
63
64       -dynamic
65
66          Display attributes of NVIDIA devices that change values as  the  GPU
67          is  working. Each of these attribute names will have a prefix string
68          at the beginning of its name. These are  FanSpeedPct ,  BoardTempC ,
69          DieTempC ,  EccErrorsSingleBit , and  EccErrorsDoubleBit .
70
71
72
73       -mixed
74
75          When displaying attribute values, assume that the machine has a het‐
76          erogeneous set of GPUs, so always include the integer value  in  the
77          prefix string .
78
79
80
81       -device <N>
82
83          Display  properties only for GPU device <N> , where <N> is the inte‐
84          ger value defined for the prefix string . Note  that  the  attribute
85          names in this output will not contain the value for <N> .
86
87
88
89       -tag string
90
91          Set  the  resource  tag  portion  of  the  intended  machine ClassAd
92          attribute  Detected<ResourceTag> to be string . If  this  option  is
93          not  specified, the resource tag is  "GPUs" , resulting in attribute
94          name  DetectedGPUs .
95
96
97
98       -prefix str
99
100          When naming attributes, use str as the prefix  string  .  When  this
101          option is not specified, the prefix string is either  CUDA or  OCL .
102
103
104
105       -simulate:D,N
106
107          For testing purposes, assume that N devices of type D were detected.
108          No discovery software is invoked. If D is 0, it refers to GeForce GT
109          330, and a default value for N is 1. If D is 1, it refers to GeForce
110          GTX 480, and a default value for N is 2.
111
112
113
114       -opencl
115
116          Prefer detection via OpenCL rather than CUDA. Without  this  option,
117          CUDA  detection  software  is  invoked first, and no further Open CL
118          software is invoked if CUDA devices are detected.
119
120
121
122       -cuda
123
124          Do only CUDA detection.
125
126
127
128       -nvcuda
129
130          For Windows platforms only, use a CUDA driver rather than  the  CUDA
131          run time.
132
133
134
135       -config
136
137          Output  in  the syntax of HTCondor configuration, instead of ClassAd
138          language. An additional  attribute  is  produced   NUM_DETECTED_GPUs
139          which is set to the number of GPUs detected.
140
141
142
143       -verbose
144
145          For  interactive  use  of the tool, output extra information to show
146          detection while in progress.
147
148
149
150       -diagnostic
151
152          Show diagnostic information, to aid in tool development.
153
154
155

Exit Status

157       condor_gpu_discovery will exit with a status value  of  0  (zero)  upon
158       success, and it will exit with the value 1 (one) upon failure.
159

Author

161       Center for High Throughput Computing, University of Wisconsin-Madison
162

164       Copyright  (C) 1990-2018 Center for High Throughput Computing, Computer
165       Sciences Department, University of Wisconsin-Madison, Madison, WI.  All
166       Rights Reserved. Licensed under the Apache License, Version 2.0.
167
168
169
170                                     datejust-man-pages/condor_gpu_discovery(1)
Impressum