awptechnologies/Arch-Dev
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

First Push

Browse Source
This commit is contained in:
awptechnologies
2024-10-30 03:27:58 -04:00
parent ba4d678cdd
commit dc910d651e
1066 changed files with 1899832 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

479
NVIDIA-Linux-x86_64-535.161.07-grid/html/powermanagement.html Normal file
View File

@@ -0,0 +1,479 @@
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta name="generator" content=
"HTML Tidy for Linux/x86 (vers 1 September 2005), see www.w3.org">
<meta http-equiv="Content-Type" content=
"text/html; charset=us-ascii">
<title>Chapter&nbsp;21.&nbsp;Configuring Power Management
Support</title>
<meta name="generator" content="DocBook XSL Stylesheets V1.68.1">
<link rel="start" href="index.html" title=
"NVIDIA Accelerated Linux Graphics Driver README and Installation Guide">
<link rel="up" href="installationandconfiguration.html" title=
"Part&nbsp;I.&nbsp;Installation and Configuration Instructions">
<link rel="prev" href="procinterface.html" title=
"Chapter&nbsp;20.&nbsp;Using the /proc File System Interface">
<link rel="next" href="dynamicpowermanagement.html" title=
"Chapter&nbsp;22.&nbsp;PCI-Express Runtime D3 (RTD3) Power Management">
</head>
<body>
<div class="navheader">
<table width="100%" summary="Navigation header">
<tr>
<th colspan="3" align="center">Chapter&nbsp;21.&nbsp;Configuring
Power Management Support</th>
</tr>
<tr>
<td width="20%" align="left"><a accesskey="p" href=
"procinterface.html">Prev</a>&nbsp;</td>
<th width="60%" align="center">Part&nbsp;I.&nbsp;Installation and
Configuration Instructions</th>
<td width="20%" align="right">&nbsp;<a accesskey="n" href=
"dynamicpowermanagement.html">Next</a></td>
</tr>
</table>
<hr></div>
<div class="chapter" lang="en">
<div class="titlepage">
<div>
<div>
<h2 class="title"><a name="powermanagement" id=
"powermanagement"></a>Chapter&nbsp;21.&nbsp;Configuring Power
Management Support</h2>
</div>
</div>
</div>
<div class="toc">
<p><b>Table of Contents</b></p>
<dl>
<dt><span class="section"><a href=
"powermanagement.html#Backgroundc73a7">Background</a></span></dt>
<dt><span class="section"><a href=
"powermanagement.html#Overview21f77">Overview</a></span></dt>
<dt><span class="section"><a href=
"powermanagement.html#PreserveAllVide719f0">Preserve all video
memory allocations</a></span></dt>
<dt><span class="section"><a href=
"powermanagement.html#SystemdConfigur74e29"><span><strong class=
"command">systemd</strong></span> Configuration</a></span></dt>
<dt><span class="section"><a href=
"powermanagement.html#ExercisingPower89cc2">Exercising power
management with <span><strong class=
"command">systemd</strong></span></a></span></dt>
<dt><span class="section"><a href=
"powermanagement.html#KnownIssuesAndWf438e">Known Issues and
Workarounds</a></span></dt>
</dl>
</div>
<div class="section" lang="en">
<div class="titlepage">
<div>
<div>
<h2 class="title" style="clear: both"><a name="Backgroundc73a7" id=
"Backgroundc73a7"></a>Background</h2>
</div>
</div>
</div>
<p>The NVIDIA Linux driver includes support for the suspend
(suspend-to-RAM) and hibernate (suspend-to-disk) system power
management operations, such as ACPI S3 and S4 on the x86_64
platform. When the system suspends or hibernates, the NVIDIA kernel
drivers prepare in-use GPUs for the sleep cycle, saving state
required to return these GPUs to normal operation when the system
is later resumed. The NVIDIA Linux driver also supports S0ix-based
s2idle system suspend (suspend-to-idle), if both the platform and
the NVIDIA GPU support it.</p>
<p>The GPU state saved by the NVIDIA kernel drivers includes
allocations made in video memory. However, these allocations are
collectively large, and typically cannot be evicted. Since the
amount of system memory available to drivers at suspend time is
often insufficient to accommodate large portions of video memory,
the NVIDIA kernel drivers are designed to act conservatively, and
normally only save essential video memory allocations.</p>
<p>The resulting loss of video memory contents is partially
compensated for by the user-space NVIDIA drivers, and by some
applications, but can lead to failures such as rendering corruption
and application crashes upon exit from power management cycles.</p>
<p>To better support power management with these types of
applications, the NVIDIA Linux driver provides a custom power
management interface intended for integration with system
management tools like <span><strong class=
"command">systemd</strong></span>. This interface is still
considered experimental. It is not used by default, but can be
taken advantage of by configuring the system as described in this
chapter.</p>
</div>
<div class="section" lang="en">
<div class="titlepage">
<div>
<div>
<h2 class="title" style="clear: both"><a name="Overview21f77" id=
"Overview21f77"></a>Overview</h2>
</div>
</div>
</div>
<p>The NVIDIA Linux driver supports the suspend and hibernate power
management operations via two different mechanisms. In this
section, each is summarized briefly with its capabilities and
requirements:</p>
<div class="variablelist">
<dl>
<dt><span class="term"><code class="computeroutput">Kernel driver
callback</code></span></dt>
<dd>
<p>When this mechanism is used, the NVIDIA kernel driver receives
callbacks from the Linux kernel to suspend, hibernate, and to
resume each GPU for which a Linux PCI driver was registered. This
is the default mechanism: it is enabled and used without explicit
configuration.</p>
<p>While this mechanism has no special requirements, yields good
results with many workloads, and has been supported by the NVIDIA
kernel driver in similar form for years, it suffers from a few
limitations. Notably, it can only preserve a relatively small
amount of video memory reliably, and it cannot support power
management when advanced CUDA features are being used.</p>
</dd>
<dt><span class="term"><code class=
"computeroutput">/proc/driver/nvidia/suspend</code></span></dt>
<dd>
<p>Instead of callbacks from the Linux kernel, this mechanism, when
used, relies on a system management tool, such as
<span><strong class="command">systemd</strong></span>, to issue
suspend, hibernate, and resume commands to the NVIDIA kernel driver
via the <code class=
"computeroutput">/proc/driver/nvidia/suspend</code> interface. It
is still considered experimental, and requires explicit
configuration to use.</p>
<p>If configured correctly, this mechanism is designed to remove
the limitations of the kernel driver callback mechanism. It
supports power management with advanced CUDA features (such as
UVM), and it is capable of saving and restoring all video memory
allocations.</p>
</dd>
</dl>
</div>
</div>
<div class="section" lang="en">
<div class="titlepage">
<div>
<div>
<h2 class="title" style="clear: both"><a name=
"PreserveAllVide719f0" id="PreserveAllVide719f0"></a>Preserve all
video memory allocations</h2>
</div>
</div>
</div>
<p>To save potentially large portions of video memory, the NVIDIA
driver supports the following two methods:</p>
<div class="variablelist">
<dl>
<dt><span class="term"><code class="computeroutput">Save
allocations in an unnamed temporary file</code></span></dt>
<dd>
<p>The NVIDIA driver uses an unnamed temporary file to save
potentially large portions of video memory. By default, this file
is created in <code class="filename">/tmp</code> during system
suspend. This location can be changed with the <code class=
"computeroutput">NVreg_TemporaryFilePath</code> nvidia.ko kernel
module parameter, e.g. <code class=
"computeroutput">NVreg_TemporaryFilePath=/run</code>. The
destination file system needs to support unnamed temporary files,
and it needs to be large enough to accommodate all the utilized
video memory copies for the duration of the power management
cycle.</p>
<p>When determining a suitable size for the video memory backing
store, it is recommended to start with the overall amount of video
memory supported by the GPUs installed in the system. For example:
<code class="computeroutput">nvidia-smi -q -d MEMORY |grep 'FB
Memory Usage' -A1</code>. Each <code class=
"computeroutput">Total</code> line returned by this command
reflects one GPU's video memory capacity, in MiB. The sum of these
numbers, plus 5% of margin, is a conservative starting point for
the size of the video memory backing store.</p>
<p>Please note that file systems such as <code class=
"filename">/tmp</code> and <code class="filename">/run</code> are
often of the type <code class="computeroutput">tmpfs</code>, and
potentially relatively small. Most commonly, the size of the type
of the file system used is controlled by <span><strong class=
"command">systemd</strong></span>. For more information, see
<a href=
"https://www.freedesktop.org/wiki/Software/systemd/APIFileSystems"
target=
"_top">https://www.freedesktop.org/wiki/Software/systemd/APIFileSystems</a>.
To achieve the best performance, file system types other than
<code class="computeroutput">tmpfs</code> are recommended at this
time.</p>
<p>Additionally, to unlock the full functionality of the interface,
the NVIDIA Linux kernel module <code class=
"computeroutput">nvidia.ko</code> needs to be loaded with the
<code class=
"computeroutput">NVreg_PreserveVideoMemoryAllocations=1</code>
module parameter. This changes the default video memory
save/restore strategy to save and restore all video memory
allocations. Also, the <code class=
"computeroutput">/proc/driver/nvidia/suspend</code> power
management mechanism (with a system management tool, such as
<span><strong class="command">systemd</strong></span>) is required
for using this interface.</p>
</dd>
<dt><span class="term"><code class="computeroutput">S0ix-based
power management</code></span></dt>
<dd>
<p>If both the platform and the NVIDIA GPU support S0ix-based power
management, then the NVIDIA Linux driver will put the GPU video
memory in self refresh mode during <span><strong class=
"command">s2idle</strong></span> system suspend. S0ix-based suspend
will consume more power than legacy S3 system suspend, but it will
enter and exit suspend/resume more quickly. Also, the
suspend/resume latency will be constant irrespective of GPU video
memory usage.</p>
<p>To check the platform S0ix state support and required
configuration, follow the steps mentioned in <a href=
"https://01.org/blogs/qwang59/2018/how-achieve-s0ix-states-linux"
target="_top">how-achieve-s0ix-states-linux</a></p>
<p>To check if the NVIDIA GPU has support for S0ix-based power
management, install the NVIDIA driver and run the following
command:</p>
<p><code class="computeroutput">grep 'Video Memory Self Refresh'
/proc/driver/nvidia/gpus/&lt;Domain&gt;:&lt;Bus&gt;:&lt;Device&gt;.0/power</code></p>
<p>For example:</p>
<p><code class="computeroutput">grep 'Video Memory Self Refresh'
/proc/driver/nvidia/gpus/0000\:01\:00.0/power</code></p>
<p>If both the platform and the GPU support S0ix-based power
management, then the S0ix support can be enabled in the NVIDIA
Linux driver by setting the <code class=
"computeroutput">nvidia.ko</code> kernel module parameter
<code class="computeroutput">NVreg_EnableS0ixPowerManagement</code>
to "1". With <code class=
"computeroutput">NVreg_EnableS0ixPowerManagement</code> set to "1"
and system suspend state set to <span><strong class=
"command">s2idle</strong></span>, the NVIDIA Linux driver will
calculate the video memory usage at system suspend time.</p>
<div class="itemizedlist">
<ul type="disc">
<li>
<p>During the S0ix suspend, if video memory usage is less than a
certain threshold, then the driver will copy video memory contents
to system memory and power off the video memory along with the GPU.
This will help in saving power.</p>
</li>
<li>
<p>During the S0ix suspend, if video memory usage is above a
certain threshold, then the video memory will be kept in
self-refresh mode while the rest of the GPU is powered down.</p>
</li>
</ul>
</div>
<p>By default, this threshold is 256 MB and it can be changed with
the <code class=
"computeroutput">NVreg_S0ixPowerManagementVideoMemoryThreshold</code>
module parameter of <code class=
"computeroutput">nvidia.ko</code>.</p>
</dd>
</dl>
</div>
<p>All the module parameters can be set on the command line when
loading the NVIDIA Linux kernel module <code class=
"computeroutput">nvidia.ko</code>, or via the distribution's kernel
module configuration files (such as those under
/etc/modprobe.d).</p>
</div>
<div class="section" lang="en">
<div class="titlepage">
<div>
<div>
<h2 class="title" style="clear: both"><a name=
"SystemdConfigur74e29" id=
"SystemdConfigur74e29"></a><span><strong class=
"command">systemd</strong></span> Configuration</h2>
</div>
</div>
</div>
<p>This section is specific to the <code class=
"filename">/proc/driver/nvidia/suspend</code> interface. This is
required if using the <code class=
"computeroutput">NVreg_PreserveVideoMemoryAllocations=1</code>
kernel module parameter or advanced CUDA features (such as UVM).
The NVIDIA Linux kernel driver requires no configuration if the
default power management mechanism is used.</p>
<p>In order to take advantage of the <code class=
"filename">/proc</code> interface, a system management tool like
<span><strong class="command">systemd</strong></span> needs to be
configured to access it at appropriate times in the power
management sequence. Specifically, the interface needs to be used
to suspend or hibernate the NVIDIA kernel drivers just before
writing to the Linux kernel's <code class=
"filename">/sys/power/state</code> interface to request entry into
the desired sleep state. The interface also needs to be used to
resume the NVIDIA kernel drivers immediately after the return from
a sleep state, as well as immediately after any unsuccessful
attempts to suspend or hibernate.</p>
<p>The following example configuration documents integration with
the <span><strong class="command">systemd</strong></span> system
and service manager, which is commonly used in modern GNU/Linux
distributions to manage system start-up and various aspects of its
operation. For systems not using <span><strong class=
"command">systemd</strong></span>, the configuration files provided
serve as a reference.</p>
<p>The <span><strong class="command">systemd</strong></span>
configuration uses the following files:</p>
<div class="variablelist">
<dl>
<dt><span class="term"><code class=
"computeroutput">/usr/lib/systemd/system/nvidia-suspend.service</code></span></dt>
<dd>
<p>A <span><strong class="command">systemd</strong></span> service
description file used to instruct the system manager to write
<code class="computeroutput">suspend</code> to the <code class=
"filename">/proc/driver/nvidia/suspend</code> interface immediately
before accessing <code class="filename">/sys/power/state</code> to
suspend the system.</p>
</dd>
<dt><span class="term"><code class=
"computeroutput">/usr/lib/systemd/system/nvidia-hibernate.service</code></span></dt>
<dd>
<p>A <span><strong class="command">systemd</strong></span> service
description file used to instruct the system manager to write
<code class="computeroutput">hibernate</code> to the <code class=
"filename">/proc/driver/nvidia/suspend</code> interface immediately
before accessing <code class="filename">/sys/power/state</code> to
hibernate the system.</p>
</dd>
<dt><span class="term"><code class=
"computeroutput">/usr/lib/systemd/system/nvidia-resume.service</code></span></dt>
<dd>
<p>A <span><strong class="command">systemd</strong></span> service
description file used to instruct the system manager to write
<code class="computeroutput">resume</code> to the <code class=
"filename">/proc/driver/nvidia/suspend</code> interface immediately
after returning from a system sleep state.</p>
</dd>
<dt><span class="term"><code class=
"computeroutput">/lib/systemd/system-sleep/nvidia</code></span></dt>
<dd>
<p>A <span><strong class="command">systemd-sleep</strong></span>
script file used to instruct the system manager to write
<code class="computeroutput">resume</code> to the <code class=
"filename">/proc/driver/nvidia/suspend</code> interface immediately
after an unsuccessful attempt to suspend or hibernate the system
via the <code class="filename">/proc/driver/nvidia/suspend</code>
interface.</p>
</dd>
<dt><span class="term"><code class=
"computeroutput">/usr/bin/nvidia-sleep.sh</code></span></dt>
<dd>
<p>A shell script used by the <span><strong class=
"command">systemd</strong></span> service description files and the
<span><strong class="command">systemd-sleep</strong></span> file to
interact with the <code class=
"filename">/proc/driver/nvidia/suspend</code> interface. The script
also manages VT switching for the X server, which is currently
needed by the NVIDIA X driver to support power management
operations.</p>
</dd>
</dl>
</div>
<p>These files are installed and enabled by nvidia-installer
automatically if systemd is detected. Installation of systemd units
can be disabled by specifying the <span><strong class=
"command">--no-systemd</strong></span> installer option.</p>
</div>
<div class="section" lang="en">
<div class="titlepage">
<div>
<div>
<h2 class="title" style="clear: both"><a name=
"ExercisingPower89cc2" id="ExercisingPower89cc2"></a>Exercising
power management with <span><strong class=
"command">systemd</strong></span></h2>
</div>
</div>
</div>
<p>This section is specific to the <code class=
"filename">/proc/driver/nvidia/suspend</code> interface, when
configured as described above. When the default power management
mechanism is used instead, or when the <code class=
"filename">/proc</code> interface is used without
<span><strong class="command">systemd</strong></span>, then the use
of <code class="computeroutput">systemctl</code> is not
required.</p>
<p>To suspend (suspend-to-RAM) or to hibernate (suspend-to-disk),
respectively, use the following commands:</p>
<div class="itemizedlist">
<ul type="disc">
<li>
<p><code class="computeroutput">sudo systemctl suspend</code></p>
</li>
<li>
<p><code class="computeroutput">sudo systemctl hibernate</code></p>
</li>
</ul>
</div>
<p>For the full list of sleep operations supported by
<span><strong class="command">systemd</strong></span>, please see
the systemd-suspend.service(8) man page.</p>
</div>
<div class="section" lang="en">
<div class="titlepage">
<div>
<div>
<h2 class="title" style="clear: both"><a name=
"KnownIssuesAndWf438e" id="KnownIssuesAndWf438e"></a>Known Issues
and Workarounds</h2>
</div>
</div>
</div>
<div class="itemizedlist">
<ul type="disc">
<li>
<p>On some systems, where the default suspend mode is <code class=
"filename">"s2idle"</code>, the system may not resume properly due
to a known timing issue in the kernel. The suspend mode can be
verified by reading the contents of the file <code class=
"filename">/sys/power/mem_sleep</code>. The following upstream
kernel changes have been proposed to fix the issue:</p>
<p><a href=
"https://lore.kernel.org/linux-pci/20190927090202.1468-1-drake@endlessm.com/"
target=
"_top">https://lore.kernel.org/linux-pci/20190927090202.1468-1-drake@endlessm.com/</a></p>
<p><a href=
"https://lore.kernel.org/linux-pci/20190821124519.71594-1-mika.westerberg@linux.intel.com/"
target=
"_top">https://lore.kernel.org/linux-pci/20190821124519.71594-1-mika.westerberg@linux.intel.com/</a></p>
<p>In the interim, the default suspend mode on the affected systems
should be set to <code class="filename">"deep"</code> using the
kernel command line parameter <code class=
"filename">"mem_sleep_default"</code> -</p>
<p><span><strong class=
"command">mem_sleep_default=deep</strong></span></p>
</li>
</ul>
</div>
</div>
</div>
<div class="navfooter">
<hr>
<table width="100%" summary="Navigation footer">
<tr>
<td width="40%" align="left"><a accesskey="p" href=
"procinterface.html">Prev</a>&nbsp;</td>
<td width="20%" align="center"><a accesskey="u" href=
"installationandconfiguration.html">Up</a></td>
<td width="40%" align="right">&nbsp;<a accesskey="n" href=
"dynamicpowermanagement.html">Next</a></td>
</tr>
<tr>
<td width="40%" align="left" valign="top">
Chapter&nbsp;20.&nbsp;Using the <code class=
"computeroutput">/proc</code> File System Interface&nbsp;</td>
<td width="20%" align="center"><a accesskey="h" href=
"index.html">Home</a></td>
<td width="40%" align="right" valign="top">
&nbsp;Chapter&nbsp;22.&nbsp;PCI-Express Runtime D3 (RTD3) Power
Management</td>
</tr>
</table>
</div>
</body>
</html>