1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
|
<!DOCTYPE html>
<html class="writer-html5" lang="en" >
<head>
<meta charset="utf-8" /><meta name="generator" content="Docutils 0.17.1: http://docutils.sourceforge.net/" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<title>Known Issues — NsightCompute 12.4 documentation</title>
<link rel="stylesheet" href="../../_static/pygments.css" type="text/css" />
<link rel="stylesheet" href="../../_static/css/theme.css" type="text/css" />
<link rel="stylesheet" href="../../_static/design-style.b7bb847fb20b106c3d81b95245e65545.min.css" type="text/css" />
<link rel="stylesheet" href="../../_static/omni-style.css" type="text/css" />
<link rel="stylesheet" href="../../_static/api-styles.css" type="text/css" />
<link rel="shortcut icon" href="../../_static/nsight-compute.ico"/>
<!--[if lt IE 9]>
<script src="../../_static/js/html5shiv.min.js"></script>
<![endif]-->
<script data-url_root="../../" id="documentation_options" src="../../_static/documentation_options.js"></script>
<script src="../../_static/jquery.js"></script>
<script src="../../_static/underscore.js"></script>
<script src="../../_static/doctools.js"></script>
<script src="../../_static/mermaid-init.js"></script>
<script src="../../_static/design-tabs.js"></script>
<script src="../../_static/version.js"></script>
<script src="../../_static/social-media.js"></script>
<script src="../../_static/js/theme.js"></script>
<link rel="index" title="Index" href="../../genindex.html" />
<link rel="search" title="Search" href="../../search.html" />
</head>
<body class="wy-body-for-nav">
<div class="wy-grid-for-nav">
<nav data-toggle="wy-nav-shift" class="wy-nav-side">
<div class="wy-side-scroll">
<div class="wy-side-nav-search" >
<a href="../../index.html">
<img src="../../_static/nsight-compute.png" class="logo" alt="Logo"/>
</a>
<div role="search">
<form id="rtd-search-form" class="wy-form" action="../../search.html" method="get">
<input type="text" name="q" placeholder="Search docs" />
<input type="hidden" name="check_keywords" value="yes" />
<input type="hidden" name="area" value="default" />
</form>
</div>
</div><div class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="Navigation menu">
<p class="caption" role="heading"><span class="caption-text">Nsight Compute</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../index.html">1. Release Notes</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../ProfilingGuide/index.html">2. Kernel Profiling Guide</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../NsightCompute/index.html">3. Nsight Compute</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../NsightComputeCli/index.html">4. Nsight Compute CLI</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Developer Interfaces</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../CustomizationGuide/index.html">1. Customization Guide</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../NvRulesAPI/index.html">2. NvRules API</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Training</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../Training/index.html">Training</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Release Information</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../Archives/index.html">Archives</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Copyright and Licenses</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../CopyrightAndLicenses/index.html">Copyright and Licenses</a></li>
</ul>
</div>
</div>
</nav>
<section data-toggle="wy-nav-shift" class="wy-nav-content-wrap"><nav class="wy-nav-top" aria-label="Mobile navigation menu" >
<i data-toggle="wy-nav-top" class="fa fa-bars"></i>
<a href="../../index.html">NsightCompute</a>
</nav>
<div class="wy-nav-content">
<div class="rst-content">
<div role="navigation" aria-label="Page navigation">
<ul class="wy-breadcrumbs">
<li><a href="../../index.html" class="icon icon-home"></a> »</li>
<li>Known Issues</li>
<li class="wy-breadcrumbs-aside">
</li>
<li class="wy-breadcrumbs-aside">
<span>v2024.1.1 |</span>
<a href="https://developer.nvidia.com/nsight-compute-history" class="reference external">Archive</a>
<span> </span>
</li>
</ul>
<hr/>
</div>
<div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article">
<div itemprop="articleBody">
<section id="known-issues">
<h1>Known Issues<a class="headerlink" href="#known-issues" title="Permalink to this headline"></a></h1>
<p><strong>Installation</strong></p>
<ul>
<li><p>The installer might not show all patch-level version numbers during installation.</p></li>
<li><p>Some command line options listed in the help of a <em>.run</em> installer of NVIDIA Nsight Compute are affecting only the archive extraction, but not the installation stage. To pass command line options to the embedded installer script, specify those options after <code class="docutils literal notranslate"><span class="pre">--</span></code> in the form of <code class="docutils literal notranslate"><span class="pre">--</span> <span class="pre">-<option></span></code>. The available options for the installer script are:</p>
<div class="highlight-text notranslate"><div class="highlight"><pre><span></span>-help : Print help message
-targetpath=<PATH> : Specify install path
-noprompt : No prompts. Implies acceptance of the EULA
</pre></div>
</div>
<p>For example, specifying only option <code class="docutils literal notranslate"><span class="pre">--quiet</span></code> extracts the installer archive without any output to the console, but still prompts for user interaction during the installation. To install NVIDIA Nsight Compute without any console output nor any user interaction, please specify <code class="docutils literal notranslate"><span class="pre">--quiet</span> <span class="pre">--</span> <span class="pre">-noprompt</span></code>.</p>
</li>
<li><p>After using the SDK Manager to install the NVIDIA Nsight Compute tools, their binary path needs to be manually added to your <code class="docutils literal notranslate"><span class="pre">PATH</span></code> environment variable.</p></li>
<li><p>See also the <a class="reference external" href="index.html#system-requirements">System Requirements</a> for more installation instructions.</p></li>
</ul>
<p><strong>Launch and Connection</strong></p>
<ul class="simple">
<li><p>Launching applications on remote targets/platforms is not supported for several combinations. See <a class="reference external" href="index.html#platform-support">Platform Support</a> for details. Manually launch the application using command line <code class="docutils literal notranslate"><span class="pre">ncu</span> <span class="pre">--mode=launch</span></code> on the remote system and connect using the UI or CLI afterwards.</p></li>
<li><p>In the NVIDIA Nsight Compute connection dialog, a remote system can only be specified for one target platform. Remove a connection from its current target platform in order to be able to add it to another.</p></li>
<li><p>Loading of CUDA sources via SSH requires that the remote connection is configured, and that the hostname/IP address of the connection matches the target (as seen in the report session details). For example, prefer my-machine.my-domain.com, instead of my-machine, even though the latter resolves to the same.</p></li>
<li><p>Other issues concerning remote connections are discussed in the documentation for <a class="reference external" href="../NsightCompute/index.html#remote-connections">remote connections</a>.</p></li>
<li><p>Local connections between NVIDIA Nsight Compute and the launched target application might not work on some ppc64le or aarch64 (sbsa) systems configured to only support IPv6. On these platforms, the <a class="reference external" href="../NsightComputeCli/index.html#environment-variables">NV_COMPUTE_PROFILER_LOCAL_CONNECTION_OVERRIDE=uds</a> environment variable can be set to use <em>Unix Domain Sockets</em> instead of <em>TCP</em> for local connections to workaround the problem. On x86_64 Linux, Unix Domain Sockets are used by default, but local TCP connections can be forced using <a class="reference external" href="../NsightComputeCli/index.html#environment-variables">NV_COMPUTE_PROFILER_LOCAL_CONNECTION_OVERRIDE=tcp</a>.</p></li>
</ul>
<p><strong>Profiling and Metrics</strong></p>
<ul class="simple">
<li><p>Profiling of 32-bit processes is not supported.</p></li>
<li><p>Profiling kernels executed on a device that is part of an SLI group is not supported. An “Unsupported GPU” error is shown in this case.</p></li>
<li><p>Profiling a kernel while other contexts are active on the same device (e.g. X server, or secondary CUDA or graphics application) can result in varying metric values for L2/FB (Device Memory) related metrics. Specifically, L2/FB traffic from non-profiled contexts cannot be excluded from the metric results. To completely avoid this issue, profile the application on a GPU without secondary contexts accessing the same device (e.g. no X server on Linux).</p></li>
<li><p>In the current release, profiling a kernel while any other GPU work is executing on the same MIG compute instance can result in varying metric values for all units. NVIDIA Nsight Compute enforces serialization of the CUDA launches within the target application to ensure those kernels do not influence each other. See <a class="reference external" href="../ProfilingGuide/index.html#serialization">Serialization</a> for more details. However, GPU work issued through other APIs in the target process or workloads created by non-target processes running simultaneously in the same MIG compute instance will influence the collected metrics. Note that it is acceptable to run CUDA processes in other MIG compute instances as they will not influence the profiled MIG compute instance.</p></li>
<li><p>On Linux kernels settings <code class="docutils literal notranslate"><span class="pre">fs.protected_regular=1</span></code> (e.g. some Ubuntu 20.04 cloud service provider instances), root users may not be able to access the <a class="reference external" href="../ProfilingGuide/index.html#serialization">inter-process lock file</a>. See the <a class="reference external" href="../ProfilingGuide/index.html#faq">FAQ</a> for workarounds.</p></li>
<li><p>Profiling only supports up to 32 device instances, including instances of MIG partitions. Profiling the 33rd or higher device instance will result in indeterminate data.</p></li>
<li><p>Enabling certain metrics can cause GPU kernels to run longer than the driver’s watchdog time-out limit. In these cases the driver will terminate the GPU kernel resulting in an application error and profiling data will not be available. Please disable the driver watchdog time out before profiling such long running CUDA kernels.</p>
<ul>
<li><p>On Linux, setting the X Config option Interactive to false is recommended.</p></li>
<li><p>For Windows, detailed information on disabling the Windows TDR is available at <a class="reference external" href="https://docs.microsoft.com/en-us/windows-hardware/drivers/display/timeout-detection-and-recovery">https://docs.microsoft.com/en-us/windows-hardware/drivers/display/timeout-detection-and-recovery</a></p></li>
</ul>
</li>
<li><p>Collecting device-level metrics, such as the NVLink metrics (<code class="docutils literal notranslate"><span class="pre">nvl*</span></code>), is not supported on <a class="reference external" href="https://www.nvidia.com/en-us/data-center/virtual-gpu-technology/">NVIDIA virtual GPUs</a> (vGPUs).</p></li>
<li><p>As of CUDA 11.4 and R470 TRD1 driver release, NVIDIA Nsight Compute is supported in a vGPU environment which requires a vGPU license. If the license is not obtained after 20 minutes, the reported performance metrics data from the GPU will be inaccurate. This is because of a feature in vGPU environment which reduces performance but retains functionality as specified <a class="reference external" href="https://docs.nvidia.com/grid/latest/grid-licensing-user-guide/index.html#software-enforcement-grid-licensing">here</a>.</p></li>
<li><p>Profiling on <a class="reference external" href="https://www.nvidia.com/en-us/data-center/virtualization/virtual-gpu-migration/">NVIDIA live-migrated virtual machines</a> is not supported and can result in undefined behavior.</p></li>
<li><p>Profiling with enabled multi-process service (MPS) can result in undefined behavior.</p></li>
<li><p>When Profiling using <em>Range Replay</em> or <em>Application Range Replay</em> with multiple CUDA Green Contexts active which belong to the same device context, the range result will contain counter values aggregated on all Green Contexts</p></li>
<li><p>The NVLink Topology section is not supported for a configuration using NVSwitch.</p></li>
<li><p>NVIDIA Nsight Compute does not support per-NVLink metrics.</p></li>
<li><p>NVIDIA Nsight Compute does not support the <em>Logical NVLink Throughput</em> table.</p></li>
<li><p>Setting a reduced NvLink Bandwidth mode does not impact the reported peak values for NvLink metrics. All peak values and corresponding percentages are calculated off the non-reduced NvLink bandwidth. Reconfiguring the NvLink Bandwidth mode using <cite>nvidia-smi`</cite> while profiling may lead to undefined tools’ behavior.</p></li>
</ul>
<ul class="simple">
<li><p>Profiling CUDA graph kernel nodes that can launch device graphs or are part of device-launchable graphs is not supported. Use <a class="reference external" href="../NsightComputeCli/index.html#command-line-options-profile">Graph Profiling</a> mode instead.</p></li>
<li><p>Profiling in <a class="reference external" href="../NsightComputeCli/index.html#command-line-options-profile">Graph Profiling</a> mode is performed on the context that is specified by the stream handle for the graph launch. Consequently, only memory allocated on this context is saved off and restored during graph replay and only kernel nodes executed on this context are profiled.</p></li>
</ul>
<ul class="simple">
<li><p>On CUDA drivers older than 530.x, profiling on Windows Subsystem for Linux (WSL) is not supported if the system has multiple physical NVIDIA GPUs. This is not affected by setting <code class="docutils literal notranslate"><span class="pre">CUDA_VISIBLE_DEVICES</span></code>.</p></li>
<li><p>Collecting software counters through PerfWorks currently forces all functions in the module of the profiled kernel to be loaded. This increases the host and device memory footprint of the target application for the remainder of the process lifetime.</p></li>
<li><p>PM Sampling is not supported when collecting a Profile Series.</p></li>
</ul>
<p><strong>Compatibility</strong></p>
<ul class="simple">
<li><p>Applications calling blocking functions on std input/output streams can result in the profiler to stop, until the blocking function call is resolved.</p></li>
<li><p>NVIDIA Nsight Compute can hang on applications using RAPIDS in versions 0.6 and 0.7, due to an issue in cuDF.</p></li>
<li><p>Profiling child processes launched via <code class="docutils literal notranslate"><span class="pre">clone()</span></code> is not supported.</p></li>
<li><p>Profiling child processes launched from Python using <code class="docutils literal notranslate"><span class="pre">os.system()</span></code> is not supported on ppc64le.</p></li>
<li><p>Profiling of Cooperative Groups kernels launched with <code class="docutils literal notranslate"><span class="pre">cuLaunchCooperativeKernelMultiDevice</span></code> is not yet supported.</p></li>
<li><p>On Linux systems, when profiling <em>bsd-csh</em> scripts, the original application output will not be printed. As a workaround, use a different C-shell, e.g. <em>tcsh</em>.</p></li>
<li><p>Attempting to use the <code class="docutils literal notranslate"><span class="pre">--clock-control</span></code> option to set the GPU clocks will fail when profiling on a GPU partition. Please use <code class="docutils literal notranslate"><span class="pre">nvidia-smi</span></code> (installed with NVIDIA display driver) to control the clocks for the entire GPU. This will require administrative privileges when the GPU is partitioned.</p></li>
<li><p>On Linux aarch64, NVIDIA Nsight Compute does not work if the <em>HOME</em> environment variable is not set.</p></li>
<li><p>NVIDIA Nsight Compute versions 2020.1.0 to 2020.2.1 are not compatible with CUDA driver version 460+ if the application launches Cooperative Groups kernels. Profiling will fail with error “UnknownError”.</p></li>
<li><p>Collecting CPU call stack information on Windows Server 2016 can hang NVIDIA Nsight Compute in some cases. Currently, the only workaround is to skip CPU call stack collection on such systems by not specifying the option <code class="docutils literal notranslate"><span class="pre">--call-stack</span></code>.</p></li>
<li><p>When profiling a script, <code class="docutils literal notranslate"><span class="pre">--target-processes</span> <span class="pre">all</span></code> may target utility executables such as <em>xargs</em>, <em>uname</em> or <em>ls</em>. To avoid profiling these, use the <code class="docutils literal notranslate"><span class="pre">--target-processes-filter</span></code> option accordingly.</p></li>
<li><p>On mobile platforms, <code class="docutils literal notranslate"><span class="pre">--kill</span></code> option is not supported with application replay mode.</p></li>
<li><p>NVIDIA Nsight Compute might show invalid characters for Unicode names and paths on Windows 10. As a workaround, use a third-party terminal emulator, e.g. Git bash.</p></li>
</ul>
<p><strong>User Interface</strong></p>
<ul class="simple">
<li><p>The API Statistics filter in NVIDIA Nsight Compute does not support units.</p></li>
<li><p>File size is the only property considered when resolving source files. Timestamps are currently ignored.</p></li>
<li><p>Terminating or disconnecting an application in the <em>Interactive Profiling</em> activity while the API Stream View is updated can lead to a crash.</p></li>
<li><p>See the <a class="reference external" href="index.html#library-support-optix">OptiX library support section</a> for limitations concerning the <a class="reference external" href="../NsightCompute/index.html#as-viewer">Acceleration Structure Viewer</a>.</p></li>
<li><p>After updating from a previous version of NVIDIA Nsight Compute on Linux, the file load dialog may not allow column resizing and sorting. As a workaround, the <em>~/.config/QtProject.conf</em> file can be edited to remove the <em>treeViewHeader</em> entry from the <em>[FileDialog]</em> section.</p></li>
</ul>
</section>
</div>
</div>
<footer>
<hr/>
<div role="contentinfo">
<p>© Copyright 2018-2024, NVIDIA Corporation & Affiliates. All rights reserved.
<span class="lastupdated">Last updated on Mar 06, 2024.
</span></p>
</div>
</footer>
</div>
</div>
</section>
</div>
<script>
jQuery(function () {
SphinxRtdTheme.Navigation.enable(true);
});
</script>
</body>
</html>
|
