BkNVR390Quadro_Vista QUADRO Win10 Win8 Win7 Release Notes 391.33

User Manual: Pdf QUADRO win10-win8-win7 release-notes 391.33

Open the PDF directly: View PDF PDF.
Page Count: 41

Release 390 Quadro, NVS,
Tesla, & Notebook Drivers -
Version 391.33
RN-07954-391.33_v01 | March 27, 2018
Windows 7, Windows 8, Windows 8.1, & Windows 10
Release Notes
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
i
Chapter : <SmallCaps>
1 Introduction to Release Notes .................................................. 1
Structure of the Document ........................................................ 1
Changes in this Edition............................................................. 1
2 Release 390 Driver Changes..................................................... 2
Version 391.33 Highlights .......................................................... 3
Existing Support.................................................................. 3
What’s New in Version 391.33.................................................. 3
What’s New in Release 390..................................................... 3
Limitations Under Windows 10................................................. 4
Discontinued and Unsupported Features..................................... 4
Advanced Instructions for this Release .......................................... 7
Turning Off V-Sync to Boost Performance.................................... 7
NVIDIA Application Configuration Engine (ACE)............................. 7
Using the WDDM Driver Model with Tesla GPU GOMs....................... 8
SLI Multi-OS – GPU Assignment in System Virtualization..................10
Changes in Version 391.33 ........................................................13
Security Updates in this Release .............................................13
Changes in Version 391.03 ........................................................14
Fixed Issues in this Release....................................................14
Changes in Version 390.77 ........................................................15
Fixed Issues in this Release....................................................15
Changes in Version 390.65 ........................................................16
NVIDIA Software Security Updates ...........................................16
Fixed Issues in this Release....................................................16
Open Issues in Version 391.33....................................................17
Windows 10 Issues ..............................................................17
Windows 7 Issues................................................................17
Windows 8 Issues................................................................17
Issues Not Caused by NVIDIA Drivers ............................................18
Windows 7 Considerations.....................................................18
Application Issues...............................................................18
Known Product Limitations .......................................................19
System Should not Enter Sleep/Hibernate Mode With SDI Capture in
Progress ..........................................................................19
Once Installed, the SDI Capture Driver Release Must Always Match the
Graphics Driver..................................................................20
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
ii
Chapter : <SmallCaps>
Some APIs do not Report Total Available Graphics Memory Correctly.. 20
Using HDMI/DisplayPort Audio with Displays that have a High Native
Resolution........................................................................ 21
Using HDMI/DisplayPort Audio in Dualview or Clone Mode Configurations22
GPU Runs at a High Performance Level (full clock speeds) in Multi-display
Modes ............................................................................ 22
Aero Must be Enabled for Windowed SLI AFR Mode Under Windows 7 . 22
SLI Connector Requirement on NVIDIA Quadro SLI Cards................. 22
Applying Workstation Application Profiles .................................. 23
3 The Release 390 Driver..........................................................24
Supported Operating Systems.................................................... 24
Supported NVIDIA Workstation GPUs............................................ 24
NVIDIA Quadro & NVS Product ............................................... 25
NVIDIA Quadro Sync Products ............................................... 27
NVIDIA Quadro SDI Products ............................................... 27
NVIDIA Quadro Blade/Embedded Graphics Board ...................... 27
NVIDIA Tesla Products .......................................................... 28
Supported NVIDIA Notebook GPUs............................................... 29
Supported Languages.............................................................. 30
Driver Installation.................................................................. 31
Minimum Hard Disk Space ..................................................... 31
Before You Begin................................................................ 31
Installation Instructions........................................................ 31
Appendix A: NVIDIA Tesla Compute Cluster Mode............................33
About Tesla Compute Cluster Mode ............................................. 33
TCC Overview ................................................................... 33
Running CUDA Applications.................................................... 34
Operating on Systems with non-TCC NVIDIA GPUs............................ 35
Setting TCC Mode .................................................................. 35
LIST OF TABLES
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
iii
Table 1 GOM and Driver Model Compatibility.............................................. 9
Table 2.1 NVIDIA Software Security Updates..............................................13
Table 3.1 Supported NVIDIA Quadro & NVS Products ....................................25
Table 3.2 Supported NVIDIA Quadro Sync II Products....................................26
Table 3.3 Supported NVIDIA Quadro Sync Products ......................................27
Table 3.4 Supported NVIDIA Quadro SDI Products........................................ 27
Table 3.5 Supported NVIDIA Quadro Blade/Embedded Graphics Board Series.......27
Table 3.6 Supported NVIDIA Tesla Products ...............................................28
Table 3.7 NVIDIA NVS Notebook GPU Support.............................................29
Table 3.8 NVIDIA Quadro Notebook GPU Support......................................... 29
Table A.1 Default TCC Mode.................................................................35
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
1
01 INTRODUCTION TO RELEASE NOTES
This edition of Release Notes describes the Release 390 family of NVIDIA® Quadro®,
NVSTM, Tesla®, and Quadro Notebook Drivers for Microsoft® Windows® 7 and later1.
NVIDIA provides these notes to describe performance improvements and bug fixes in
each documented version of the driver.
Structure of the Document
This document is organized in the following sections:
“Release 390 Driver Changes” on page 2 gives a summary of changes, and fixed and
open issues in this version.
“The Release 390 Driver” on page 24 describes the NVIDIA products and languages
supported by this driver, the system requirements, and how to install the driver.
“NVIDIA Tesla Compute Cluster Mode” on page 33 describes the Tesla Compute
Cluster mode.
Changes in this Edition
This edition of the Release Notes for Windows 7 and later includes information about
NVIDIA graphics driver version 391.33, and lists changes made to the driver since
version 388.16. These changes are discussed beginning with the chapter “Release 390
Driver Changes” on page 2.
1. Includes Windows 7, Windows 8, Windows 8.1, and Windows 10.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
2
02 RELEASE 390 DRIVER CHANGES
This chapter describes open issues for version 391.33, and resolved issues and driver
enhancements for versions of the Release 390 driver up to version 391.33. The chapter
contains these sections:
“Version 391.33 Highlights” on page 3
“Advanced Instructions for this Release” on page 7
“Changes in Version 391.33” on page 13
“Changes in Version 391.03” on page 14
“Changes in Version 390.77” on page 15
“Changes in Version 390.65” on page 16
“Open Issues in Version 391.33” on page 17
“Issues Not Caused by NVIDIA Drivers” on page 18
“Known Product Limitations” on page 19
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
3
Chapter 02 : R
ELEASE
390 D
RIVER
C
HANGES
Version 391.33 Highlights
This section provides highlights of version 391.33 of the NVIDIA Release 390 Driver for
Windows 7 and later.
Existing Support
What’s New in Version 391.33
What’s New in Release 390
Discontinued and Unsupported Features
Limitations in This Release
Existing Support
This release supports the following APIs:
•Open Computing Language (OpenCLTM software) 1.2 for NVIDIA® KeplerTMand
later GPUs
•OpenGL® 4.6
•Vulkan® 1.0
•DirectX 11
•DirectX 12 (Windows 10)
•NVIDIA® CUDA® 9.1
This driver installs nView Desktop Manager version 148.92.
What’s New in Version 391.33
This driver adds security updates for driver components. See “Security Updates in
this Release” on page 13 for specifics.
Added support for the NVIDIA Quadro GV100 (64-bit Windows OS only).
See also What’s New in Release 390 and“Changes in Version 391.33” on page 13.
What’s New in Release 390
Quadro ODE Driver
Release 390 is an 'Optimal Drivers for Enterprise'(ODE) branch release. ODE branches
are dedicated to relatively long term stability for ISV certification, OEMs, and Enterprise
customers.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
4
Chapter 02 : R
ELEASE
390 D
RIVER
C
HANGES
New Features
Added support for CUDA 9.1
Added support for Windows 10 Fall Creators Update
Added DP Inband Stereo option to the NVIDIA Control Panel ->Manage 3D Settings-
>Stereo-Display Mode setting.
Limitations Under Windows 10
Cross-adapter Clone Mode
Under Windows 10, clone mode across graphics processors is handled by the Windows
OS, and not by the NVIDIA driver. Consequently, there are no controls for cross-adapter
clone mode in the NVIDIA Control Panel under Windows 10.
Functionality of some display controls in the NVIDIA Control Panel may be affected if
cross-adapter clone mode is enabled through the Windows Control Panel. In this case,
use the Windows Control Panel to adjust display settings.
Optimus Systems
Because cross-adapter clone mode is handled by the Windows 10 OS, there are no display
controls in the NVIDIA Control Panel if no display is connected to the NVIDIA GPU.
Discontinued and Unsupported Features
Advance Notification of Future Discontinued Support
32-bit Operating Systems
After Release 390, NVIDIA will no longer release drivers for 32-bit operating systems
for any GPU architecture.
NVS 310/315
After Release 390, NVIDIA will no longer support the following NVS products:
•NVS 315
•NVS 310
Quad-Buffered Stereo
After Release 390, the NVIDIA Professional Drivers for Windows will no longer
support select quad-buffered stereo features. See the NVIDIA Knowledge Base Article
for details.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
5
Chapter 02 : R
ELEASE
390 D
RIVER
C
HANGES
Discontinued Support in this Release
Discontinued Support for Select Quadro Products
Beginning with the Release 378, the NVIDIA professional drivers no longer support
the following NVIDIA Quadro products:
The NVIDIA support team will continue to address critical driver issues for these
products in Release 375 through March 1, 2018. However, future driver enhancements
and optimizations in driver releases after Release 375 will not support these products.
Limitations in This Release
The following features are not currently supported or have limited support in this driver
release:
External Graphics
External GPU Surprise Removal
Not all applications have been designed to address surprise removal of the external
GPU; disconnection of the external GPU while applications are running is not
advised.
Mixed GeForce/Quadro Products
Mixed GeForce/Quadro products are supported (Geforce GPU + Quadro eGPU, or
Quadro GPU + Geforce eGPU), but requires installation of the GeForce driver
package. The Quadro package does not install GeForce drivers.
Deep Color (10 bpc) on HDR-Capable Monitors Under Windows 10 RS2 and RS3
Depending on the monitors you have connected, Windows may initially set the color
depth to 8-bpc (bits per channel). You can then enable deep color (10bpc) through the
NVIDIA Control Panel. If you enable deep color on HDR-capable monitors, the screen
may become blank due to the current SDR/HDR (standard dynamic range/high dynamic
range) management policies in Windows 10 RS2 and RS3. In this event, the driver will
revert back to the 8-bpc color space.
Quadro Quadro Blade/Embedded Quadro Notebook
Quadro Plex 7000 Quadro 4000M Quadro 5010M
Quadro 6000 Quadro 3000M Quadro 5000M
Quadro 5000 Quadro 100M Quadro 4000M
Quadro 4000 Quadro 500M Quadro 3000M
Quadro 2000 Quadro 2000M
Quadro 2000 D Quadro 1000M
Quadro 600
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
6
Chapter 02 : R
ELEASE
390 D
RIVER
C
HANGES
Video Memory Support
For Windows 7 64-bit, this driver recognizes up to the total available video memory on
Quadro cards for Direct3D and OpenGL applications.
For Windows 7 32-bit, this driver recognizes only up to 4 GB of video memory on Quadro
cards for DirectX, OpenGL, and CUDA applications.
Experimental OpenCL 2.0 Features
Select features in OpenCL 2.0 are available in the driver for evaluation purposes only. The
following are the features as well as a description of known issues with these features in
the driver:
Device side enqueue
•The current implementation is limited to 64-bit platforms only.
•OpenCL 2.0 allows kernels to be enqueued with
global_work_size
larger than the
compute capability of the NVIDIA GPU. The current implementation supports only
combinations of
global_work_size
and
local_work_size
that are within the compute
capability of the NVIDIA GPU.
The maximum supported CUDA grid and block size of NVIDIA GPUs is available
at http://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#compute-
capabilities. For a given grid dimension, the
global_work_size
can be determined by
CUDA grid size x CUDA block size.
•For executing kernels (whether from the host or the device), OpenCL 2.0 supports
non-uniform ND-ranges where
global_work_size
does not need to be divisible by
the
local_work_size
. This capability is not yet supported in the NVIDIA driver, and
therefore not supported for device side kernel enqueues.
Shared virtual memory
•The current implementation of shared virtual memory is limited to 64-bit platforms
only.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
7
Chapter 02 : R
ELEASE
390 D
RIVER
C
HANGES
Advanced Instructions for this Release
This section clarifies instructions for successfully accomplishing the following tasks:
Turning Off V-Sync to Boost Performance
NVIDIA Application Configuration Engine (ACE)
Using the WDDM Driver Model with Tesla GPU GOMs
SLI Multi-OS – GPU Assignment in System Virtualization
Turning Off V-Sync to Boost Performance
To get the best benchmark and application performance measurements, turn V-Sync off
as follows:
1Open the NVIDIA Control Panel and make sure that Advanced Settings is selected from
the control panel tool bar.
2From the Select a Task pane, under 3D Settings, click Manage 3D Settings, then click
the Global Settings tab.
3From the Global presets pull-down menu, select Base profile.
4From the Settings list box, select Vertical sync and change its value to Force off, then
click Apply.
5From the Global presets pull-down menu, select 3D App - Default Global Settings
(the driver’s default profile) or use the application profile that matches the application
you are testing, then click Apply.
Be sure to close the NVIDIA Control Panel completely —leaving it open will affect
benchmark and application performance.
NVIDIA Application Configuration Engine (ACE)
This driver includes the NVIDIA Application Configuration Engine (ACE), which
automatically detects the workstation application and configures the appropriate profile
settings in the NVIDIA Control Panel.
See the NVIDIA Quadro Professional Drivers: NVIDIA Control Panel Quick Start Guide for
more information about this feature.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
8
Chapter 02 : R
ELEASE
390 D
RIVER
C
HANGES
Using the WDDM Driver Model with Tesla GPU GOMs
Tesla GPU Operation Modes
Tesla K20X and K20 GPU Accelerators offer the ability to specify the GPU operation
mode using NV-SMI/NVML. (refer to: https://developer.nvidia.com/nvidia-
management-library-nvml).
By setting the GPU operation mode, developers can selectively turn off certain features in
the GPU to get the best performance per watt for certain workloads.
The following are the supported GOMs:
Compute-Only: For running compute tasks only.
By default the Tesla K20X and K20 accelerator boards ship in this mode.
Low-Double Precision: For graphics applications that don’t require high bandwidth
double precision.
This is recommended for workloads that are not sensitive to double precision but at
the same time need graphics capabilities.
All On: This is recommended only when the workload needs full double precision as
well as graphics capabilities.
WDDM and TCC Driver Models
Along with the GPU operation mode, the developer needs to select the compatible driver
model for the Tesla K20X and K20.
Tesla Compute Cluster (TCC): Optimized for running compute workloads.
Windows Device Driver Model (WDDM): Designed for graphics application and not
recommended for compute workloads.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
9
Chapter 02 : R
ELEASE
390 D
RIVER
C
HANGES
Compatibility Between GOM and Driver Models
Table 1 shows which GPU operation modes are compatible with which driver models.
The compute-only GOM is supported only on the TCC driver model, while the WDDM
driver model supports only GOM modes that enable graphics.
The compute-only GOM and WDDM are incompatible and should not be used
simultaneously.
The Tesla K20 Active Accelerators for workstations ship in “compute-only” mode and
cannot be modified. Therefore, use only the TCC driver model with these products.
Using the WDDM Driver Model
To use the WDDM driver model with Tesla K20x and K20 GPU Accelerators for servers,
first switch the GOM mode from compute-only to All On, then switch from TCC to
WDDM.
Do not attempt to specify the driver model by editing the registry. Doing so can result in
compute-only GOM and WDDM being configured simultaneously, which might require
a clean installation of the driver to fix.
Always use NVIDIA-provided tools to specify a processing mode or to switch between
driver models. Such tools include nvidia-smi or the NVIDIA Control Panel->Manage
Maximus Settings page. These tools provide warnings in the case of a conflict.
Table 1 GOM and Driver Model Compatibility
GOM TCC Driver
Model WDDM Driver
Model Use Case Support
All On YES YES All use cases are supported.
Compute-Only YES NO The following are
unsupported:
X11 and those that require
X11 (GLInterop, OCL
conformance and VIPER)
32-bit Windows OS
Low Double Precision YES YES All use cases supported.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
10
Chapter 02 : R
ELEASE
390 D
RIVER
C
HANGES
SLI Multi-OS – GPU Assignment in System Virtualization
On systems with two or more graphics cards installed, this driver supports a hypervisor's
ability to directly assign GPUs to guest virtual machines (VMs). This direct assignment
allows each guest VM to run on their own operating system with their own GPU and
driver. The assignment allows full GPU performance and functionality in the guest VM.
Hardware Platform Requirements
To make use of GPU passthrough with virtual machines running Windows and Linux,
the hardware platform must support the following features:
A CPU with hardware-assisted instruction set virtualization: Intel VT-x or AMD-V.
Platform support for I/O DMA remapping.
On Intel platforms the DMA remapper technology is called Intel VT-d.
On AMD platforms it is called AMD IOMMU.
Support for these feature varies by processor family, product, and system, and should be
verified at the manufacturer's website.
Supported Hypervisors
The following hypervisors are supported:
Supported Graphics Cards
The following GPUs are supported for device passthrough:
Hypervisor Notes
Citrix XenServer Version 6.0 and later.
VMware vSphere (ESX / ESXi) Version 5.1 and later.
Parallels Workstation Extreme Version 4 and later
GPU Family Boards supported
Volta Quadro: GV100
Tesla: V100
Pascal Quadro: P2000, P4000, P5000, P6000, GP100
Tesla: P100, P40, P4
Maxwell Quadro: K2200, M2000, M4000, M5000, M6000, M6000 24GB
Tesla: M60, M6
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
11
Chapter 02 : R
ELEASE
390 D
RIVER
C
HANGES
Notes and Known Issues
VMware
•PCI I/O hole may need to be changed for Windows 64-bit VMs.
Windows 64-bit VMs may require that you edit the VM configuration file to
configure a larger PCI I/O hole for the GPU.
•Access Control Services is required on some switches.
Starting with ESX 5.0 Update 1, Access Control Services (ACS) is required on any
switches in the PCIe hierarchy above a PCIe device that is to be used for
passthrough. If ACS is not present, ESX will not allow the device to be assigned
directly to a VM.
To allow assignment of devices below switches that do not support ACS, you can
disable the ACS check as follows:
esxcfg-advcfg -k true disableACSCheck
•Configuring passthrough to a Windows Server 2008R2 VM.
The VMware WDDM driver must be manually installed on Windows Server 2008
R2.
Follow the steps provided at http://communities.vmware.com/message/
1423263#1425288 .
•MSI translation must be disabled for ESXi 5.1 and 5.5.
Blue-screen crashes may occur on VMs with assigned GPUs if MSI is initially
enabled for passthrough devices.
To avoid this issue, disable MSI translation by setting
pciPassthru0.msiEnabled =
"FALSE"
in the VM’s VMX file.
Disabling MSI translation is not needed for ESXi 6.0 and later.
Citrix Xenserver
•XenServer 6.1 workaround for 64-bit MMIO failures on PCI passthrough.
Dom0 kernel may relocate GPU BARs to an invalid location immediately beyond
the end of physical RAM. There are two workaround options to address this
condition:
Kepler Quadro: K2000, K4000, K4200, K5000, K5200, K6000
Tesla: K10, K20, K20x, K20Xm, K20c, K20s, K40m, K40c,
K40s, K40st, K40t, K80
Fermi Tesla: C2050, C2075, M2050, M2070, M2070Q, M2075, M2090
GPU Family Boards supported
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
12
Chapter 02 : R
ELEASE
390 D
RIVER
C
HANGES
Workaround option #1: Add ‘
pci=use_crs
’ to the kernel parameters by running the
following command in Dom0:
/opt/xensource/libexec/xen-cmdline --set-dom0 pci=use_crs
Workaround option #2: Install the hotfix provided by Citrix at http://
support.citrix.com/article/CTX137645):
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
13
Changes in Version 391.33
Changes in Version 391.33
The following sections list the important changes and the most common issues resolved
since driver version 391.03.
Security Updates in this Release
For more information about medium to high-severity issues, refer to the NVIDIA
Product Security page.
Table 2.1 NVIDIA Software Security Updates
CVE ID NVIDIA Issue
Number Description
CVE-2018-6247 2003798 Vulnerability in the kernel mode layer (nvlddmkm.sys)
handler for DxgkDdiEscape where a NULL pointer
dereference may lead to Denial of Service or Escalation of
Privileges.
CVE-2018-6248 1970661
1969787 Vulnerability in the kernel mode layer handler for
DxgkDdiEscape where the software uses a sequential
operation to read or write a buffer, but it uses an
incorrect length value that causes it to access memory
that is outside of the bounds of the buffer which may lead
to denial of service or possible escalation of privileges.
CVE-2018-6249 1960097
1952160
2050960
Vulnerability in kernel mode layer handler where a NULL
pointer dereference may lead to denial of service or
potential escalation of privileges.
CVE-2018-6250 200337413 Vulnerability in the kernel mode layer (nvlddmkm.sys)
handler for DxgkDdiEscape where a NULL pointer
dereference occurs which may lead to denial of service or
possible escalation of privileges.
CVE-2018-6251 2050736 Vulnerability in Directx 10 Usermode driver, where
specially crafted pixel shader can cause writing to
unallocated memory which may lead to denial of service
or potential code execution.
CVE-2018-6252 1893760 Vulnerability in the kernel mode layer handler for
DxgkDdiEscape where the software allows an actor access
to restricted functionality that is unnecessary to
production usage, and which may result in denial of
service.
CVE-2018-6253 2060652 Vulnerability in DirectX and OpenGL Usermode drivers
where specially crafted pixel shader can cause infinite
recursion which may lead to denial of service.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
14
Changes in Version 391.03
Changes in Version 391.03
The following sections list the important changes and the most common issues resolved
since driver version 390.77.
Fixed Issues in this Release
[Multi-GPU Mosaic]: Displays go blank upon disabling the Mosaic topology, then the
primary GPU is disabled upon restarting the system.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
15
Changes in Version 390.77
Changes in Version 390.77
The following sections list the important changes and the most common issues resolved
since driver version 390.65.
Fixed Issues in this Release
[Quadro 4000M][Workstation Stereo]: Black shadows appear around stereo images
after exiting and then re-launching the OpenGL application.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
16
Changes in Version 390.65
Changes in Version 390.65
The following sections list the important changes and the most common issues resolved
since driver version 388.16.
NVIDIA Software Security Updates
For more information about NVIDIA's vulnerability management, visit the NVIDIA
Product Security page.
Fixed Issues in this Release
[Multi-GPU Mosaic][G-Sync][Quadro GP100+ Nvlink]: Mosaic topologies cannot be
set.
CVE ID NVIDIA Issue
Number Description
CVE-2017-5753 2034846 Security Bulletin: NVIDIA Driver Security Updates for CPU
Speculative Side Channel Vulnerabilities
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
17
Open Issues in Version 391.33
Open Issues in Version 391.33
As with every released driver, version 391.33 of the Release 390 driver has open issues
and enhancement requests associated with it. This section includes lists of issues that are
either not fixed or not implemented in this version. Some problems listed may not have
been thoroughly investigated and, in fact, may not be NVIDIA issues. Others may have
workaround solutions.
Windows 10 Issues
[Assimilate Scratch]: The application may crash due to a kernel exception in the
NVIDIA OpenGL driver.
If you encounter this issue, roll back the driver to version 385.90, or disable Deep Color within
the application as follows:
From the startup screen choose System Settings.
- In the System Settings Dialog, choose the Advanced tab.
- From the list (under Graphics) select "OpenGL: Disable 'deep color'
- Click the boolean button to turn the setting on.
[Quadro M6000][Quadro Sync]: Unexpected flashing may appear on the display
connected to the second GPU.
Please consult your NVIDIA support team for details on resolving this behavior.
[Multi-GPU Mosaic][Quadro Sync]: Application corruption and possible blue-screen
crash occurs when enabling WARP while Mosaic and Quadro Sync are enabled.
Windows 7 Issues
[Quadro M6000][Quadro Sync]: Unexpected flashing may appear on the display
connected to the second GPU.
Please consult your NVIDIA support team for details on resolving this behavior.
[Quadro Sync]: Frame overruns occur on low resolution channels when using Quadro
Sync swap synchronization.
Windows 8 Issues
[Quadro M6000][Quadro Sync]: Unexpected flashing may appear on the display
connected to the second GPU.
Please consult your NVIDIA support team for details on resolving this behavior.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
18
Issues Not Caused by NVIDIA Drivers
Issues Not Caused by NVIDIA Drivers
This section lists issues that are not due to the NVIDIA driver as well as features that are
not meant to be supported by the NVIDIA driver for Windows 7.
“Windows 7 Considerations” on page 18
“Application Issues” on page 18
Windows 7 Considerations
Windows DWM Disabled for SLI Mosaic Mode
Due to compatibility issues, when using SLI Mosaic mode the driver turns off the
Windows 7 Desktop Window Manager (DWM). As a result, DWM-managed desktop
features such as Windows Aero or Windows Flip 3D will not be available.
Limitation
When switching the refresh rate from 59 Hz to 60Hz, the refresh rate remains at 59 Hz.
See the Microsoft KB article KB2006076 at http://support.microsoft.com/kb/2006076.
Application Issues
[3ds Max 2017]: DirectX 9, Mode Texture of some models disappears after some time if
no operation (such as zoom, pan, or rotate) is performed on the model.
[Quadro K2200][Bunkspeed Shot 2014]: Render Argument exception occurs in
Accurate Mode, and Internal Rendering Error occurs with Fast Mode.
Quadro K2200/K60 are not supported by Bunkspeed software.
Quadro 6000: OpenCL/OpenGL interoperability performance suffers with
clEnqueueReleaseGLObjects.
To avoid this issue, use CUDA.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
19
Known Product Limitations
This section describes problems that will not be fixed. Usually, the source of the problem
is beyond the control of NVIDIA. Following is the list of problems and where they are
discussed in this document:
“System Should not Enter Sleep/Hibernate Mode With SDI Capture in Progress” on
page 19
“Once Installed, the SDI Capture Driver Release Must Always Match the Graphics
Driver” on page 20
“Some APIs do not Report Total Available Graphics Memory Correctly” on page 20
“Using HDMI/DisplayPort Audio with Displays that have a High Native Resolution”
on page 21
“Using HDMI/DisplayPort Audio in Dualview or Clone Mode Configurations” on
page 22
“GPU Runs at a High Performance Level (full clock speeds) in Multi-display Modes”
on page 22
“Aero Must be Enabled for Windowed SLI AFR Mode Under Windows 7” on page 22
“SLI Connector Requirement on NVIDIA Quadro SLI Cards” on page 22
“Applying Workstation Application Profiles” on page 23
System Should not Enter Sleep/Hibernate Mode With SDI
Capture in Progress
SDI capture programs will become non-responsive upon resume from sleep or hibernate
modes.
This is not a typical use case, so power management tests should not be performed when
testing SDI.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
20
Once Installed, the SDI Capture Driver Release Must Always
Match the Graphics Driver
To use the Quadro SDI Capture card, you need to install the NVIDIA WDM driver for the
Quadro SDI Capture card after installing the Quadro professional driver. The WDM
driver version must be from the same branch as the installed NVIDIA graphics driver.
If at a later time you install a different Quadro professional driver version–for example,
when upgrading from a Release 260 driver to a Release 265 driver–you must reinstall the
WDM driver from the same driver branch as well– in this example, Release 265.
You must install the updated WDM driver even if remove the SDI Capture card and
uninstall the WDM driver. Portions of the old WDM driver remain on the system, and the
resulting mismatch with the updated graphics driver results in a faulty driver
installation. Installing a matching WDM driver corrects this issue.
Some APIs do not Report Total Available Graphics Memory
Correctly
Background–TAG Memory
In the Windows Display Driver Model (WDDM), Total Available Graphics (TAG)
memory is reported as the sum of
•Dedicated Video Memory (video memory dedicated for graphics use)
•Dedicated System Memory (system memory dedicated for graphics use), and
•Shared System Memory (system memory shared between the graphics subsystem
and the CPU).
The values for each of these components are computed according to WDDM guidelines
when the NVIDIA Display Driver is loaded.
Issue
NVIDIA has found that some TAG-reporting APIs represent video memory using 32-bits
instead of 64-bits, and consequently do not properly report available graphics memory
when the TAG would otherwise exceed 4 gigabytes (GB). This results in under reporting
of available memory and potentially undesirable behavior of applications that rely on
these APIs to report available memory.
The reported memory can be severely reduced. For example, 6 GB might be reported as
454 MB, and 8 GB might be reported as 1259 MB.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
21
NVIDIA Action for Some GeForce-based Systems
For GeForce GPUs with 2.75 GB or less of video memory, the NVIDIA display driver
constrains TAG memory to just below 4 GB1. In this scenario, the Shared System Memory
component of TAG is limited first, before limiting Dedicated Video Memory.
This is a policy decision within the driver, and results in reliable reporting of sub-4 GB
TAG memory.
When TAG Reporting Would Not Be Limited
For GeForce-based GPUs with more than 2.75 GB of video memory, as well as all Quadro
and Tesla GPUs, the NVIDIA display driver does not constrain TAG memory reporting.
The disadvantage of constraining TAG on systems with larger amounts of video and
system memory is that memory which otherwise would be available for graphics use is
no longer available. Since shared system memory is limited first, driver components and
algorithms utilizing shared system memory may suffer performance degradation when
TAG is constrained.
Since these and similar scenarios are prevalent in many Workstation applications, the
NVIDIA driver avoids constraining TAG on all Quadro and Tesla-based systems.
Likewise, the driver does not constrain TAG for GeForce-based systems with more than
2.75 GB of video memory.
Using HDMI/DisplayPort Audio with Displays that have a
High Native Resolution
To use HDMI/DisplayPort audio with some displays that have a native resolution higher than
1920x1080, you must set the display to a lower HD resolution.
Some HDMI TV's have a native resolution that exceeds the maximum supported HD
mode. For example, TVs with a native resolution of 1920x1200 exceed the maximum
supported HD mode of 1920x1080.
Applying this native mode results in display overscan which cannot be resized using the
NVIDIA Control Panel since the mode is not an HD mode.
To avoid this situation and provide a better user experience, the driver treats certain
TVs– such as the Viewsonic VX2835wm and the Westinghouse LVM- 37w3–as a DVI
monitor when applying the native mode. Because the driver does not treat the TV as an
HDMI in this case, the HDMI audio is not used.
1. The WDDM guidelines dictate minimum and maximum values for the components, but the dis-
play driver may further constrain the values that are reported (within the allowed minimum and
maximum).
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
22
Using HDMI/DisplayPort Audio in Dualview or Clone Mode
Configurations
Two Audio-enabled Ports
In a multi-display configuration where both HDMI/DisplayPort audio ports are enabled,
only the primary display will provide the audio.
One Audio-enabled Port
In a multi-display configuration where only one audio port is enabled, such as when one
display is a DVI display, then the HDMI/DisplayPort display can provide the audio
whether is it the primary or secondary display.
GPU Runs at a High Performance Level (full clock speeds) in
Multi-display Modes
This is a hardware limitation and not a software bug. Even when no 3D programs are
running, the driver will operate the GPU at a high performance level in order to
efficiently drive multiple displays. In the case of SLI or multi-GPU PCs, the second GPU
will always operate with full clock speeds; again, in order to efficiently drive multiple
displays. Today, all hardware from all GPU vendors have this limitation.
Aero Must be Enabled for Windowed SLI AFR Mode Under
Windows 7
Windows 7 Aero must be enabled in order to achieve SLI acceleration using windowed
AFR mode.
SLI Connector Requirement on NVIDIA Quadro SLI Cards
The SLI connector that links two SLI cards is needed for proper SLI operation. However,
the connector can be removed if you do not intend to enable SLI mode. If you remove the
connector, then you must make sure that SLI mode is disabled from the NVIDIA control
panel. Enabling SLI mode without the SLI connector installed will result in video
corruption.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
23
Applying Workstation Application Profiles
Background
The workstation application profiles are software settings used by the NVIDIA
Display Drivers to provide optimum performance when using a selected application.
The profile also works around known application issues and bugs.
If there is an available setting for an application, it should be used, otherwise incorrect
behavior or reduced performance is likely to occur.
Issues
Configuration changes require that you restart the application.
Once an application is running, it does not receive notification of configuration
changes. Therefore, if you change the configuration while the application is running,
you must exit and restart the application for the configuration changes to take effect.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
24
03 THE RELEASE 390 DRIVER
This chapter covers the following main topics:
“Supported Operating Systems” on page 24
“Supported NVIDIA Workstation GPUs” on page 24
“Supported NVIDIA Notebook GPUs” on page 29
“Supported Languages” on page 30
“Driver Installation” on page 31
Supported Operating Systems
The Release 390 driver, version 391.33, has been tested with
Microsoft Windows® 10, and supports both 32-bit and 64-bit versions.
Microsoft Windows® 8.1, and supports both 32-bit and 64-bit versions.
Microsoft Windows® 8, and supports both 32-bit and 64-bit versions.
Microsoft Windows® 7, and supports both 32-bit and 64-bit versions.
Supported NVIDIA Workstation GPUs
The following tables list the NVIDIA products supported by the Release 390 driver,
version 391.33.
Table 3.1, “Supported NVIDIA Quadro & NVS Products” on page 25
Table 3.2, “Supported NVIDIA Quadro Sync II Products” on page 26
Table 3.3, “Supported NVIDIA Quadro Sync Products” on page 27
Table 3.4, “Supported NVIDIA Quadro SDI Products” on page 27
Table 3.5, “Supported NVIDIA Quadro Blade/Embedded Graphics Board Series” on
page 27
Table 3.6, “Supported NVIDIA Tesla Products” on page 28
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
25
Chapter 03 : T
HE
R
ELEASE
390 D
RIVER
NVIDIA Quadro & NVS Product
Table 3.1 Supported NVIDIA Quadro & NVS Products
Product GPU Architecture Notes
NVIDIA Quadro GV100 Volta 64-bit Windows OS only
NVIDIA Quadro GP100 Pascal
NVIDIA Quadro P6000 Pascal
NVIDIA Quadro P5000 Pascal
NVIDIA Quadro P4000 Pascal
NVIDIA Quadro P2000 Pascal
NVIDIA Quadro P1000 Pascal
NVIDIA Quadro P600 Pascal
NVIDIA Quadro P400 Pascal
NVIDIA Quadro M6000 24GB Maxwell
NVIDIA Quadro M6000 Maxwell
NVIDIA Quadro M5000 Maxwell
NVIDIA Quadro M4000 Maxwell
NVIDIA Quadro M2000 Maxwell
NVIDIA Quadro K6000 Kepler
NVIDIA Quadro K5200 Kepler
NVIDIA Quadro K5000 Kepler
NVIDIA Quadro K4200 Kepler
NVIDIA Quadro K4000 Kepler
NVIDIA Quadro K2200 Maxwell
NVIDIA Quadro K2000 Kepler
NVIDIA Quadro K1200 Maxwell
NVIDIA Quadro K620 Maxwell
NVIDIA Quadro K600 Kepler
NVIDIA Quadro K420 Kepler
NVIDIA Quadro 410 Kepler
NVIDIA NVS 810 Maxwell
NVIDIA NVS 510 Kepler
NVIDIA NVS 315 Fermi
NVIDIA NVS 310 Fermi
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
26
Chapter 03 : T
HE
R
ELEASE
390 D
RIVER
Table 3.2 Supported NVIDIA Quadro Sync II Products
Product GPU Architecture
NVIDIA Quadro GP100 Pascal
NVIDIA Quadro P6000 Pascal
NVIDIA Quadro P5000 Pascal
NVIDIA Quadro P4000 Pascal
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
27
Chapter 03 : T
HE
R
ELEASE
390 D
RIVER
NVIDIA Quadro Sync Products
NVIDIA Quadro SDI Products
NVIDIA Quadro Blade/Embedded Graphics Board
Table 3.3 Supported NVIDIA Quadro Sync Products
Product GPU Architecture
NVIDIA Quadro M6000 24GB Maxwell
NVIDIA Quadro M6000 Maxwell
NVIDIA Quadro M5000 Maxwell
NVIDIA Quadro M4000 Maxwell
NVIDIA Quadro K6000 Kepler
NVIDIA Quadro K5200 Kepler
NVIDIA Quadro K5000 Kepler
NVIDIA Quadro K4200 Kepler
Table 3.4 Supported NVIDIA Quadro SDI Products
Product GPU Architecture
NVIDIA Quadro K6000 Kepler
NVIDIA Quadro K5200 Kepler
NVIDIA Quadro K5000 Kepler
NVIDIA Quadro K4200 Kepler
NVIDIA Quadro K4000 Kepler
Note: Quadro SDI products are not supported under Windows 8.
Table 3.5 Supported NVIDIA Quadro Blade/Embedded Graphics Board
Series
Product GPU Architecture
NVIDIA Quadro P5000 Pascal
NVIDIA Quadro P3000 Pascal
NVIDIA Quadro M5000 SE Maxwell
NVIDIA Quadro M3000 SE Maxwell
NVIDIA Quadro K3100M Kepler
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
28
Chapter 03 : T
HE
R
ELEASE
390 D
RIVER
NVIDIA Tesla Products
The Tesla driver package is designed for systems that have one or more Tesla products
installed.
Only one GHIC can be connected to the server in a Tesla system.
This release of the Tesla driver supports CUDA C/C++ applications and libraries that
rely on the CUDA C Runtime and/or CUDA Driver API.
Table 3.6 Supported NVIDIA Tesla Products
Product GPU Architecture Notes
Tesla P-Series Products
NVIDIA Tesla P100 Pascal
NVIDIA Tesla P40 Pascal
NVIDIA Tesla P4 Pascal
Tesla M-Series Products
NVIDIA Tesla M60 Maxwell
NVIDIA Tesla M6 Maxwell
Tesla K-Series Products
NVIDIA Tesla K80 Kepler 64-bit OS only
NVIDIA Tesla K40(m/c/s/st/t) Kepler 64-bit OS only
NVIDIA Tesla K20(x/c/m/Xm/s) Kepler 64-bit OS only
NVIDIA Tesla K10 Kepler
Tesla S-Class Products
NVIDIA Tesla S2050 Fermi
Tesla M-Class Products
NVIDIA Tesla M2090 Fermi
NVIDIA Tesla M2075 Fermi
NVIDIA Tesla M2070 Fermi
NVIDIA Tesla M2070Q Fermi
NVIDIA Tesla M2050 Fermi
Tesla C-Class Products
NVIDIA Tesla C2075 Fermi
NVIDIA Tesla C2070 Fermi
NVIDIA Tesla C2050 Fermi
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
29
Chapter 03 : T
HE
R
ELEASE
390 D
RIVER
Supported NVIDIA Notebook GPUs
The notebook driver is part of the NVIDIA Verde Notebook Driver Program, and can be
installed on supported NVIDIA notebook GPUs. However, please note that your
notebook original equipment manufacturer (OEM) provides certified drivers for your
specific notebook on their website. NVIDIA recommends that you check with your
notebook OEM about recommended software updates for your notebook. OEMs may not
provide technical support for issues that arise from the use of this driver.
The following tables list the NVIDIA notebook products supported by the Release 390
driver, version 391.33:
Table 3.7 NVIDIA NVS Notebook GPU Support
Notebook Products GPU Architecture
NVS 5400M
NVS 5200M
NVS 4200M
Fermi
Fermi
Fermi
Table 3.8 NVIDIA Quadro Notebook GPU Support
Notebook Products GPU Architecture
Quadro P5000
Quadro P4000
Quadro P3000
Quadro M5500
Quadro M5000M
Quadro M4000M
Quadro M3000M
Quadro M2000M
Quadro M1000M
Quadro M600M
Quadro M500M
Pascal
Pascal
Pascal
Maxwell
Maxwell
Maxwell
Maxwell
Maxwell
Maxwell
Maxwell
Maxwell
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
30
Chapter 03 : T
HE
R
ELEASE
390 D
RIVER
Supported Languages
The Release 390 Graphics Drivers supports the following languages in the main driver
Control Panel:
Quadro K5000M
Quadro K5100M
Quadro K4000M
Quadro K4100M
Quadro K3000M
Quadro K3100M
Quadro K2000M
Quadro K2200M
Quadro K2100M
Quadro K1000M
Quadro K1100M
Quadro K620M
Quadro K610M
Quadro K510M
Quadro K500M
Kepler
Kepler
Kepler
Kepler
Kepler
Kepler
Kepler
Maxwell
Kepler
Kepler
Kepler
Maxwell
Kepler
Kepler
Kepler
English (USA) German Portuguese (Euro/
Iberian)
English (UK) Greek Russian
Arabic Hebrew Slovak
Chinese (Simplified) Hungarian Slovenian
Chinese (Traditional) Italian Spanish
Czech Japanese Spanish (Latin America)
Danish Korean Swedish
Dutch Norwegian Thai
Finnish Polish Turkish
French Portuguese (Brazil)
Table 3.8 NVIDIA Quadro Notebook GPU Support (continued)
Notebook Products GPU Architecture
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
31
Chapter 03 : T
HE
R
ELEASE
390 D
RIVER
Driver Installation
Minimum Hard Disk Space
The hard disk space requirement for international is minimum 1158 MB.
The hard disk space requirement for English-only is minimum 570 MB.
Before You Begin
nTune
If you have previously installed NVIDIA nTune, NVIDIA recommends that you uninstall
nTune before installing this driver. After the driver install is complete, you can reinstall
NVIDIA nTune.
Notebooks
Check to make sure that your notebook has a supported GPU (see “Supported
NVIDIA Notebook GPUs” on page 29).
It is recommended that you back up your current system configuration.
If you own a Dell Inspiron 1420, Dell XPS M1330, or Dell XPS M1530, or Dell
LatitudeD630 or D630c, it is highly recommended that you first install this Dell
software update.
SLI Mosaic Mode
You must make sure SLI Mosaic mode is disabled before installing a new driver over a
previously installed driver. If SLI Mosaic mode is active on your displays when you
install the new driver, the driver will not install properly.
Installation Instructions
1Follow the instructions on the NVIDIA .com Web site driver download page to locate
the appropriate driver to download, based on your hardware and operating system.
2Click the driver download link.
The license agreement dialog box appears.
3Click Accept if you accept the terms of the agreement, then either open the file or save
the file to your PC and open it later.
4Open the NVIDIA driver installation .EXE file to launch the NVIDIA InstallShield
Wizard.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
32
Chapter 03 : T
HE
R
ELEASE
390 D
RIVER
5Follow the instructions in the NVIDIA InstallShield Wizard to complete the
installation.
Note: If you are overinstalling the driver (installing over a previous
driver without first removing the previous driver), then you must
reboot your computer in order to complete the installation.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
33
APPENDIX A NVIDIA TESLA COMPUTE
CLUSTER MODE
This chapter describes the Tesla Compute Cluster (TCC) mode.
“About Tesla Compute Cluster Mode” on page 33
“Operating on Systems with non-TCC NVIDIA GPUs” on page 35
“Setting TCC Mode” on page 35
About Tesla Compute Cluster Mode
TCC Overview
Tesla Compute Cluster (TCC) mode is designed for compute cluster nodes that have one
or more Tesla or supported Quadro products installed.
Benefits
TCC drivers make it possible to use NVIDIA GPUs in nodes with non-NVIDIA
integrated graphics.
NVIDIA GPUs on systems running the TCC drivers will be available via Remote
Desktop, both directly and via cluster management systems that rely on Remote
Desktop.
NVIDIA GPUs will be available to applications running as a Windows service (i.e. in
Session 0) on systems running the Tesla driver in TCC mode.
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
34
Appendix A : NVIDIA T
ESLA
C
OMPUTE
C
LUSTER
M
ODE
TCC Does not Support Graphics Acceleration
TCC mode does not provide CUDA–DirectX/OpenGL interoperability.
It is a “non-display” driver, and NVIDIA GPUs using this driver will not support
DirectX or OpenGL hardware acceleration.
Running CUDA Applications
This release of the Tesla/Quadro driver supports CUDA C/C++ applications and
libraries that rely on the CUDA C Runtime and/or CUDA Driver API.
NVIDIA GPUs running the Tesla/Quadro driver in TCC mode will be available for
CUDA applications running via services or Remote Desktop.
In this release, all GPUs will be in compute exclusive mode. As a result, only one
CUDA context may exist on a particular device at a time.
SDK applications that use graphics will not run properly under TCC mode. The
following are examples of CUDA SDK applications that are not supported:
bicubicTexture boxFilter cudaDecodeD3D9 smokeParticles
cudaDecodeGL fluidsD3D9 fluidsGL SobelFilter
imageDenoising Mandelbrot marchingCubes volumeRender
nbody oceanFFT particles
postProcessGL recursiveGaussian simpleD3D10
simpleD3D10Texture simpleD3D11Texture simpleD3D9
simpleD3D9Texture simpleGL simpleTexture3D
Release 390 Quadro, NVS, Tesla, & Notebook Drivers - Version 391.33 RN-07954-391.33_v01
|
35
Appendix A : NVIDIA T
ESLA
C
OMPUTE
C
LUSTER
M
ODE
Operating on Systems with non-TCC NVIDIA GPUs
NVIDIA GPUs running under TCC mode may coexist with other display devices.
The Tesla/Quadro driver is overinstalled over any NVIDIA display driver in the
system–the NVIDIA Tesla driver then becomes the only driver for all NVIDIA GPUs
in the system.
If the Tesla/Quadro driver is uninstalled at a later time, the previous driver is not
restored.
NVIDIA GPUs that do not support TCC mode will appear as “VGA adapters” in the
Windows Device Manager and can be used to drive displays.
Non-supported NVIDIA GPUs can still function as CUDA devices, but the GPU’s
graphics functionality is not available to applications.
Setting TCC Mode
To change the TCC mode, use the NVIDIA smi utility as follows:
nvidia-smi -g (GPU ID) -dm (0 for WDDM, 1 for TCC)
The following table shows the default TCC/non-TCC mode for supported products.
Table A.1 Default TCC Mode
Product Default Mode
K20Xm/c TCC
K20m TCC Off
C2075 TCC
C2050 TCC Off
C2070 TCC Off
S2050 TCC
M2050 TCC
M2070 TCC
M2075 TCC
M2070-Q TCC Off
M2090 TCC
Quadro (Kepler/Maxwell) TCC Off
www.nvidia.com
Notice
ALL NVIDIA DESIGN SPECIFICATIONS, REFERENCE BOARDS, FILES, DRAWINGS, DIAGNOSTICS, LISTS, AND
OTHER DOCUMENTS (TOGETHER AND SEPARATELY, "MATERIALS") ARE BEING PROVIDED "AS IS." NVIDIA MAKES
NO WARRANTIES, EXPRESSED, IMPLIED, STATUTORY, OR OTHERWISE WITH RESPECT TO THE MATERIALS, AND
EXPRESSLY DISCLAIMS ALL IMPLIED WARRANTIES OF NONINFRINGEMENT, MERCHANTABILITY, AND FITNESS
FOR A PARTICULAR PURPOSE.
Information furnished is believed to be accurate and reliable. However, NVIDIA Corporation assumes no
responsibility for the consequences of use of such information or for any infringement of patents or other
rights of third parties that may result from its use. No license is granted by implication of otherwise under
any patent rights of NVIDIA Corporation. Specifications mentioned in this publication are subject to change
without notice. This publication supersedes and replaces all other information previously supplied. NVIDIA
Corporation products are not authorized as critical components in life support devices or systems without
express written approval of NVIDIA Corporation.
Macrovision Compliance Statement
NVIDIA Products that are Macrovision enabled can only be sold or distributed to buyers with a valid and
existing authorization from Macrovision to purchase and incorporate the device into buyer's products.
Macrovision copy protection technology is protected by U.S. patent numbers 5,583,936; 6,516,132;
6,836,549; and 7,050,698 and other intellectual property rights. The use of Macrovision's copy protection
technology in the device must be authorized by Macrovision and is intended for home and other limited
pay-per-view uses only, unless otherwise authorized in writing by Macrovision. Reverse engineering or
disassembly is prohibited.
Third Party Notice
Portions of the NVIDIA system software contain components licensed from third parties under the following
terms:
Clang & LLVM:
Copyright (c) 2003-2015 University of Illinois at Urbana-Champaign.
All rights reserved.
Portions of LLVM's System library:
Copyright (C) 2004 eXtensible Systems, Inc.
Developed by:
LLVM Team
University of Illinois at Urbana-Champaign
http://llvm.org
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and
associated documentation files (the "Software"), to deal with the Software without restriction, including
without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following
conditions:
* Redistributions of source code must retain the above copyright notice, this list of conditions and the
following disclaimers.
* Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the
following disclaimers in the documentation and/or other materials provided with the distribution.
www.nvidia.com
* Neither the names of the LLVM Team, University of Illinois at Urbana-Champaign, nor the names of its
contributors may be used to endorse or promote products derived from this Software without specific prior
written permission.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING
BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
NONINFRINGEMENT. IN NO EVENT SHALL THE CONTRIBUTORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS WITH
THE SOFTWARE.
Vulkan Notice
Copyright (c) 2015-2016 The Khronos Group Inc.
Copyright (c) 2015-2016 LunarG, Inc.
Copyright (c) 2015-2016 Valve Corporation
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and/or
associated documentation files (the "Materials"), to deal in the Materials without restriction, including
without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
of the Materials, and to permit persons to whom the Materials are furnished to do so, subject to the
following conditions:
The above copyright notice(s) and this permission notice shall be included in all copies or substantial
portions of the Materials.
THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING
BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
NONINFRINGEMENT.
IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
CONNECTION WITH THE MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS.
Unless otherwise noted in the LICENSE.txt file in the install folder, all components of the Vulkan Runtime
are licensed under the above license. Licenses for any components not so licensed are listed in the
LICENSE.txt file.
Trademarks
NVIDIA, the NVIDIA logo, NVIDIA nForce, GeForce, NVIDIA Quadro, are registered trademarks or trademarks
of NVIDIA Corporation in the United States and/or other countries.
HDMI, the HDMI logo, and High-Definition Multimedia Interface are trademarks or registered trademarks of
HDMI Licensing LLC.
OpenGL® and the oval logo are trademarks or registered trademarks of Silicon Graphics, Inc. in the United
States and/or other countries worldwide. Additional license details are available on the SGI website.
OpenCL and the OpenCL logo are trademarks of Apple Inc. used by permission by Khronos.
Vulkan and the Vulkan logo are trademarks of the Khronos Group Inc.
Other company and product names may be trademarks or registered trademarks of the respective owners
with which they are associated.
Copyright
© 2018 NVIDIA Corporation. All rights reserved.

Navigation menu