SGI® Virtu VN200 Graphics Node Hardware User’s Guide 007-5483-001
COPYRIGHT © 2002-2008 SGI. All rights reserved; provided portions may be copyright in third parties, as indicated elsewhere herein. No permission is granted to copy, distribute, or create derivative works from the contents of this electronic documentation in any manner, in whole or in part, without the prior written permission of SGI.
Record of Revision 007-5483-001 Version Description 001 April, 2008. Original printing.
iv 007-5483-001
Contents About This Guide Audience 1 . . . . . . . . . . . . . . . . . . . . . . xiii Structure of this Guide . . . . . . . . . . . . . . . . . . xiii Related Publications . . . . . . . . . . . . . . . . . . . xiv Conventions . . . . . . . . . . . . . . . . . . . . . xv Product Support . . . . . . . . . . . . . . . . . . . . xv Reader Comments . . . . . . . . . . . . . . . . . . .
Regulatory Specifications . . . . . . . . . . . . . . . . . . 17 Server Model Number . . . . . . . . . . . . . . . . . 17 CE Notice and Manufacturer's Declaration of Conformity . . . . . . . 18 Electromagnetic Emissions . . . . . . . . . . . . . . . . 18 FCC Notice (USA Only) . . . . . . . . . . . . . . . . 18 Industry Canada Notice (Canada Only) . . . . . . . . . . . 19 VCCI Class A Notice (Japan Only) . . . .
3 System Procedures Connecting a Digital Monitor . . . . . . . . . . . . . . . . 43 Connecting a Keyboard and Mouse . . . . . . . . . . . . . . . 44 Configuring Firmware and Drivers . Power On/Off the Cluster . . . . . . . . . . . . . . 45 . . . . . . . . . . . . . . . . . 45 Powering Off Manually . . . . . . . . . . . . . . . . . 46 Verifying the Graphics Card Installation for Linux . . . . . . . . . .
Hardware Replacement Procedures . Before You Begin . . . . . . . . . . . . . . . . 64 . . . . . . . . . . . . . . . . 64 Tools and Supplies Needed . . . . . . . . . . . . . . . . 64 Removing the Front Bezel . . . . . . . . . . . . . . . 64 Removing the Graphics Node from the Rack . . . . . . . . . . . 65 Removing the Side Cover. . . . . . . . . . . . . . . . . . 66 Replacing the Power Supplies . . . . . .
Figures Figure 1-1 Figure 1-2 Figure 1-3 Figure 1-4 Figure 1-5 Figure 1-6 Figure 1-7 Figure 2-1 Figure 2-2 Figure 2-3 Figure 2-4 Figure 2-5 Figure 2-6 Figure 2-7 Figure 2-8 Figure 2-9 Figure 2-10 Figure 2-11 Figure 2-12 Figure 2-13 Figure 2-14 Figure 2-15 Figure 3-1 Figure 3-2 Figure 3-3 Figure 3-4 Figure 3-5 Figure 3-6 Figure 3-7 Figure 3-8 007-5483-001 SGI Altix 1300 Cluster with Virtu VN200 Graphics Nodes. Unpacking the Rack. . . . . . . . . . . . Unpacking the Graphics Nodes . . . . . . . .
Figure 3-9 Figure 3-10 Figure 3-11 Figure 3-12 Figure 3-13 Figure 3-14 Figure 3-15 Figure 3-16 Figure 3-17 Figure 3-18 Figure 3-19 x Disk Drive Removal . . . Cooling Fans Removal . . Graphics Card Removal . . BMC Daughter Card Removal PCI Riser Board Removal . Air Baffle Removal . . . DIMM Removal . . . . Node Board Removal . . Install a CPU Heatsink . . Air Baffles . . . . . . Control Panel Removal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Tables 007-5483-001 Table 1-1 SGI Virtu VN200 System Configurations . . . . . 11 Table 1-2 Graphics Node Power Consumption. . . . . . . 15 Table 2-1 Node Board Jumper Definitions . . . . . . . 30 Table 2-2 Node Board Connector Definitions . . . . . . . 30 Table 2-3 Memory DIMM Pairs . . . . . . . . . . . 33 Table 2-5 DIMM Configurations . . . . . . . . . . . 34 Table 2-4 Memory Sizes and SGI Part Numbers . . . . . .
xii 007-5483-001
About This Guide This guide provides information about how to install, configure, administrate, and repair an SGI Virtu VN200. Note: Refer to “Related Publications” on page xiv, for a listing of relevant SGI Technical Publications that can provide more detailed information about SGI cluster head nodes, compute nodes, system software, and rack hardware. Audience This guide is intended for system administrators. Use this guide to: • Unpack and install the hardware.
Related Publications Your SGI Virtu VN200 system ships with a variety of hardware and software documents in both hard copy and soft copy formats. Hard copy documents are in the packing box and soft copy documents are located on your system hard disk in /usr/local/Factory-Install/Docs. Additional third-party documentation may be shipped on removable media (CD/DVD) included with your shipment.
Conventions The following conventions are used throughout this document: Convention Meaning Command This fixed-space font denotes literal items such as commands, files, routines, path names, signals, messages, and programming language structures. variable The italic typeface denotes variable entries and words or concepts being defined. Italic typeface is also used for book titles. user input This bold fixed-space font denotes literal items that the user enters in interactive sessions.
Reader Comments If you have comments about the technical accuracy, content, or organization of this document, contact SGI. Be sure to include the title and document number of the manual with your comments. (Online, the document number is located in the front matter of the manual. In printed manuals, the document number is located at the bottom of each page.) You can contact SGI in any of the following ways: • Send e-mail to the following address: techpubs@sgi.
Chapter 1 1. SGI Virtu VN200 Overview This chapter provides an introduction to the SGI Virtu VN200 graphics node (refer to Figure 1-1) and introduces information necessary to understand the system.
SGI Virtu VN200 Overview Product Description This document is intended as an overview of some of the common operations that system administrators may have to perform to set-up, boot, re-configure (upgrade) or troubleshoot an SGI Altix XE cluster with SGI Virtu VN200 graphics nodes.
Product Description The primary hardware components in an SGI Virtu VN200 system are: • Head node(s) (SGI Altix XE250 servers) • Compute nodes (SGI Altix XE320 servers) • Graphics nodes (SGI Virtu VN200 graphics nodes) • Network interconnect components (Gigabit Ethernet switches, InfiniBand switches, PCI cards, and cables) • System console, monitor, keyboard, and mouse The head node is connected to the interconnect network and also to the public network, typically via the local area network (LAN)
SGI Virtu VN200 Overview Warnings and Cautions Read the following safety and warnings and cautions when operating your SGI Virtu VN200 cluster. ! Warning: System power on/off: The power button DOES NOT turn off the system ! Caution: Electrostatic discharge (ESD) and ESD protection: ESD can damage disk drives, boards, and other parts. We recommend that you perform all replacement procedures on an ESD-safe worksurface.
Site Planning Site Planning Ensure that all site requirements are met before you install and boot your system. If you have questions about the site requirements, contact a site planning representative by e-mail (site@sgi.com).
SGI Virtu VN200 Overview Selecting a Final Location Consider the following issues when you select a final location for the system: • Will the system fit in its intended location? Carefully calculate the total system dimensions to ensure that it will fit in its intended final location. • Does the intended system location provide adequate access space for maintenance? • Even if the system will fit in its intended final location, you must have room to maintain it.
Unpacking the Rack Unpacking the Rack Ensure that the temperature of the rack is acclimated to the installation environment and that the system crate is stable, and in an upright position. Refer to Figure 1-2 as you follow these steps; the numbered illustrations correspond to the numbered steps. Ensure that the temperature of the rack is acclimated to the installation environment and that the system container is stable, and in an upright position.
SGI Virtu VN200 Overview ! Warning: Serious personal injury or death could result if the rack becomes unbalanced or gains too much momentum when rolled. Always get the assistance of other qualified personnel when rolling a rack. 6. Get four qualified persons to carefully roll the rack down the ramp and onto the floor. 7. Install the rear door onto rack by aligning the door on the hinge pins pushing downward. 8. Connect the door ground wire.
Unpacking the Graphics Nodes Unpacking the Graphics Nodes The SGI Virtu VN200 graphics nodes are packed as shown in Figure 1-3. Unpack each graphics node and place it on a stable ESD safe worksurface or position each graphics node on the rack shelf as shown in Figure 1-4.
SGI Virtu VN200 Overview Installing the Rack Shelf Install the rack shelf as shown in Figure 1-4. The graphics nodes require 4U of rack space. Five graphics nodes can be placed on the rack shelf. Position each graphics node on the rack shelf as shown in Figure 1-4.
System Configurations and Cabling System Configurations and Cabling Table 1-1 lists the system configurations for the SGI Virtu VN200. SGI Altix ICE clusters and SGI Altix IA64 configurations will be supported in a future release of the product.
SGI Virtu VN200 Overview Figure 1-5 shows an SGI Virtu VN200 graphics node configured with an SGI Altix XE cluster. Monitor Keyboard Mouse Base Gigabit Ethernet switch for Admin.
System Configurations and Cabling Figure 1-6 shows an SGI Virtu VN200 graphics node configured with an SGI Altix ICE cluster. This system configuration will be supported in a future product release.
SGI Virtu VN200 Overview Graphics Node Dimensions The graphics node physical characteristics are shown in Figure 1-7. The graphics node dimensions are as follows: • Height: 6.72 in. (17.07 cm) • Width: 3.44 in.(8.74 cm) • Depth: 32.44 in. (82.
Electrical Requirements Electrical Requirements The SGI Virtu VN200 accepts single-phase input voltage only. It typically connects to a single-phase or 3-phase rack power distribution unit (PDU) that outputs single-phase 220 VAC. Ensure that the required voltage is available and is within a reasonable distance of the intended location. If it is not, the site must be wired for the required voltage.
SGI Virtu VN200 Overview Environmental Requirements Electromagnetic interference (EMI), electrostatic discharge (ESD), vibration, and humidity can cause problems for computer systems. Operating Temperature 10º to 40º C (50º to 104º F) Non-operating Temperature -40º to 70º C (-40º to 158º F) Operating Relative Humidity 8% to 90% (non-condensing) Non-operating Relative Humidity 5% to 95% (non-condensing) Acoustics The acoustic sound level of the graphics node is 62.5 dBA.
Regulatory Specifications Electrostatic Discharge SGI designs and tests its products to ensure that they resist the effects of electrostatic discharge (ESD). However, it is still possible for ESD to cause problems that range from data errors and lockups to permanent component damage. To protect the systems from ESD, follow these precautions: • Minimize the use of carpeting at computer locations (or consider special static-reducing carpet). • Ensure that all electronic devices are properly grounded.
SGI Virtu VN200 Overview CE Notice and Manufacturer's Declaration of Conformity The “CE” symbol indicates compliance of the device to directives of the European Community. A “Declaration of Conformity” in accordance with the standards has been made and is available from SGI upon request. Electromagnetic Emissions This section provides the contents of electromagnetic emissions notices for various countries. FCC Notice (USA Only) This equipment complies with Part 15 of the FCC Rules.
Regulatory Specifications ! Caution: Changes or modifications to the equipment not expressly approved by the party responsible for compliance could void your authority to operate the equipment. Industry Canada Notice (Canada Only) This Class A digital apparatus meets all requirements of the Canadian Interference-Causing Equipment Regulations.
SGI Virtu VN200 Overview 20 007-5483-001
Chapter 2 2. System Features An SGI Virtu VN200 (refer to Figure 2-1) is an integrated compute environment that can scale to 288 (refer to Figure 2-2). The standard cooling method for the SGI Virtu VN200 is blower-driven air cooling. Optional supplemental water cooling will be available (in a 42U water-cooled rack) in a future release.
System Features Server Racks Each SGI Virtu VN200 rack contains the head nodes, compute nodes, an InfiniBand switch, a Gigabit Ethernet switch, a system console, power distribution units (PDUs) and graphics nodes (refer to Figure 2-2). The racks are typically air-cooled.
Virtu VN200 Graphics Node Virtu VN200 Graphics Node Each Virtu VN200 graphics node (refer to Figure 2-3) contains the hardware components listed below: • • • • • • • • • • 1 Supermicro® X7DWT-INF node board that uses the Intel 5400 Seaburg chip set (same node board used in SGI Altix XE320 servers) 2 Intel® Xeon® quad-core 5400 sequence processors with 1333- or 1600-MT/s FSB 2 to 8 800-MT/s FB-DIMMs 2 Gigabit Ethernet ports (GLAN 1 and GLAN2) 1 BMC LAN port available from GLAN 1 1 4x DDR InfiniBand interfa
System Features Front Panel Components and Status LEDs The front panel of the SGI Virtu VN200 provides status LEDs, system disk access, and power supply access (refer to Figure 2-4).
Virtu VN200 Graphics Node Rear Panel Components Figure 2-5 shows the location of the rear panel components on the SGI Virtu VN200 graphics node. The rear panel component descriptions are listed below: • • • • • • • • • • Input AC - Connects to system input AC power cord USB ports - Universal Serial Bus (USB 2.
System Features Node Board At the heart of the SGI Virtu VN200 lies a dual-processor socket node board, which is based on the Intel 5400P chipset (refer to Figure 2-6 and Figure 2-7). Memory DIMMs Processors Power connectors SATA ports Fan connectors Power connectors S I/O DIMM 3A J8B2 DIMM 2B J8B1 DIMM 2A J7B3 DIMM 1B J7B2 DIMM 1A J7B1 JWR1 PWSMB J17 20-Pin PWR FAN1/2 CPU 1 FAN3/4 Intel 5400 FAN5/6 North Bridge Battery JWOL CTRL 4-Pin Aux.
Node Board Node Board Features Refer to the node board block diagram, Figure 2-8. CPU • Dual Intel® 64-bit Xeon LGA 771 quad-core/dual-core processors at a front side bus speed of 1.600 GHz or 1.333 GHz Memory • Eight 240-pin DIMM sockets with support up to 32 GB DDR2 FBD (fully buffered) ECC 800 MT/s memory Chipset • Intel 5400 chipset, including: the 5400 Memory Control Hub (MCH), and the Enterprise South Bridge 2 (ESB2) Expansion Slots • One PCI-Express (Gen. 2.
System Features ACPI Features • • • Slow blinking LED for suspend state indicator Main switch override mechanism ACPI Power Management Onboard I/O • • • • • • • • Four SATA2 system disk ports (only two ports are used) One SIMSO IPMI slot One Intel Gigabit Ethernet 82563EB controller supports two Gigabit LAN ports One COM port Up to four Universal Serial Bus 2.
Node Board PROCESSOR#2 PROCESSOR#1 1333/1600 MT/S PCI-EXP x16 FBD CHNL2 MCH FBD CHNL3 PCI-EXP x8 FBD DIMM FBD CHNL1 FBD DIMM Intel 5400 FBD DIMM FBD CHNL0 FBD DIMM PCI-E SLOT 1333/1600 MT/S VGA CONN VGA ES1000 PCI 33MHz RJ45 GB LAN GILGAL USB 2.0 LPC KUMERAN RJ45 3.
System Features Jumper Quick-reference Refer to Figure 2-7 on page 26 when using Table 2-1 and Table 2-2.
Node Board Table 2-2 Node Board Connector Definitions (continued) Connector Description JWOL Wake-on LAN header USB 0/1 back panel USB 0/1 for external use (JUSB1) USB 2/3 front panel USB 2/3 VGA VGA Connector (JVGA1) SIMSO SIMSO IPMI slot SATA 0-3 Intel SATA 0-3 connectors LED Indicator description LE1 Onboard Power LED Indicator LE2 InfiniBand Link LED (green) LE3 InfiniBand Activity LED (yellow) Memory Overview Each node board supports 2 two to eight 240-pin DIMMs that can provid
System Features Slot 2 DIMM 1B DIMM 2B DIMM 3B DIMM 4B Slot 1 DIMM 1A DIMM 2A DIMM 3A DIMM 4A Channel 1 Channel 0 Channel 2 Branch 0 Channel 3 Branch 1 MCH Figure 2-9 Memory Block Diagram DIMM 2B DIMM 3A DIMM 3B DIMM 2A DIMM 4A DIMM 1B DIMM 4B DIMM 1A Figure 2-10 32 Node Board DIMM Layout 007-5483-001
Node Board Memory DIMM Pairs Table 2-3 lists the memory DIMM pairs for the SGI Virtu VN200. Table 2-3 Memory DIMM Pairs DIMM Pair DIMM Slot Pair 1 1A and 2A Pair 2 3A and 4A Pair 3 1B and 2B Pair 4 3B and 4B Memory Configuration Rules • You can install a minimum of 2 and a maximum of 8 FBDIMMs on the node board. • You must install FB-DIMMs in pairs (refer to Table 2-3). • Both FB-DIMMs within a pair must have the same capacity and speed.
System Features Table 2-4 Memory Sizes and SGI Part Numbers Memory Size and Type Speed SGI Part Number 1-GB FB-DIMM 800 MT/s 9010132 2-GB FB-DIMM 800 MT/s 9010133 4-GB FB-DIMM 800 MT/s 9010134 Table 2-5 lists the DIMM configurations for the SGI Virtu VN200.
Node Board AOC-SIMSO (BMC) Card This section describes the AOC-SIMSO baseboard management controller (BMC). The intelligent platform management interface (IPMI) management utility defines the protocols used for interfacing with the BMC on the node board. Use the GLAN 1 port to connect to the BMC interface (refer to Figure 2-11).
System Features IPMI controller enables you to communicate with the BMC via a Java-based web interface.
Node Board Power Supplies Each hot-pluggable power supply connects to the power distribution assembly in the center of the chassis. The power supplies are located in the chassis as shown in Figure 2-12. The power supply specifications are listed in Table 2-6.
System Features Cooling Fans Each power supply has built in cooling fans that pull cool air in from the front of the chassis. Two chassis cooling fans pull cool air in from the front of the chassis and blow it across the node board heatsinks and memory DIMMs. The chassis cooling fans receive power from the node board assembly (refer to Figure 2-6). The graphics card assembly has its own cooling fan and heatsink assembly.
Node Board System Disk Drives The system disk drive are controlled by the SATA ports on the node board (refer to Figure 2-7 on page 26. The following system disk drive options are supported by the SGI Virtu VN200 (refer to Figure 2-14): • • • 2.5-in, 80 GB, 7200 RPM SATA 2.5-in, 120 GB, 7200 RPM SATA 2.5-in, 160 GB, 7200 RPM SATA Note: Shut down the operating system before replacing a disk drive. Disk drives can be replaced while the system is powered on (hot-plugged).
System Features Graphics Cards The following graphics cards are supported on the SGI Virtu VN200. Other cards may be supported in a future release. • • • • NVIDIA Quadro FX1700 NVIDIA Quadro FX3700 NVIDIA Quadro FX4600 NVIDIA Quadro FX5600 Figure 2-15 Graphics Card Table 2-7 lists the SGI Virtu VN200 graphics card specifications. Table 2-7 40 Graphics Card Specifications Type Memory Memory Bandwidth Performance Display Connectors NVIDIA Quadro FX1700 512 MB DDR2 12.
Node Board Accelerated Driver Installation Accelerated graphics drivers from NVIDIA are preinstalled by SGI on the SGI Virtu VN200 graphics node. These drivers are not part of the standard SLES or RHEL distribution of Linux. A copy of these drivers is available on the SGI Propack CD and/or supportfolio. SGI distributes these drivers as .rpm files. SGI provides the limited support of the factory installed NVIDIA graphics drivers.
System Features 42 007-5483-001
Chapter 3 3. System Procedures Connecting a Digital Monitor The NVIDIA Quadro FX card supports two DVI-I outputs. The DVI-I output farthest away from the stereo connector is the primary channel for the system. By default this port outputs an analog signal when the system is not configured. If you attach digital monitor to the system, you can reset the DVI-I port by performing the following steps. 1. Attach the monitor. 2. Log into the node remotely as root. 3.
System Procedures Connecting a Keyboard and Mouse A keyboard/mouse is not supplied as standard for the system but may be provided as an option from SGI. Connect a USB keyboard and a USB mouse to your system, as shown in Figure 3-1. If your keyboard/mouse is further than 10 feet (3 meters) away from the USB ports on the system, then optional USB extension cables are required. Monitor Keyboard Mouse Base Gigabit Ethernet switch for Admin.
Configuring Firmware and Drivers Configuring Firmware and Drivers The SGI Virtu VN200 uses the same BIOS images as the SGI Altix XE320 server, with customized settings. The SGI Virtu VN200 also uses factory-installed, accelerated NVIDIA graphics drivers and X.org windowing system; SGI provides limited support for these SGI qualified NVIDIA graphics drivers.
System Procedures If your cluster uses the Scali Manage administrative software (release 5.6.x) or later, you can power on/off specific nodes or the entire system using the graphical user interface. Select Management Menu -> Power Mgt ->Power Off. The compute nodes can be halted from the Scali Manage GUI by selecting the nodes and choosing “halt system” and “power down” from the System Management menu. A command line interface is also available to power-on/off or check status.
Verifying the Graphics Card Installation for Linux Verifying the Graphics Card Installation for Linux To verify the graphics card is installed and seen correctly by Linux do the following steps. From a Linux root prompt enter: # /sbin/lspci | grep VGA Check the output for information similar to the following: 01:00.0 VGA compatible controller: nVidia Corporation G80 [Quadro FX xxx] (rev xx) 08:01.0 VGA compatible controller: ATI Technologies Inc ES1000 (rev 02) 01:00.0 is the primary card for the system.
System Procedures Starting the X Window System The SGI Virtu VN200 requires a number of system-specific X Windows System configuration settings for various configurations. The following section provides an overview on how to how to customize the xorg.conf file. The X Window System is the only window system provided for the SGI Virtu VN200 running Linux. The SGI Virtu VN200 uses an X server (Xorg) from the opensource xorg project (www.x.org).
Starting the X Window System Documentation NVIDIA driver documentation is located in /usr/share/doc/NVIDIA_GLX-1.0/README.txt. Refer to the man pages for the commands listed above and the X.Org website at: http://www.x.org. Start Graphics The init 5 command start the system in the graphical mode. When init 5 is used, the graphics console displays a login screen.
System Procedures xorg.conf Configuration File Xorg.conf is the file containing configuration information for the default X server. It is best to think about the organization of the xorg.conf file as tree (refer to Figure 3-2).
Starting the X Window System The xorg.conf file is located in the /usr/etc/X11 directory. Make a copy of the factory default xorg.conf file so that you can make your configuration changes. An example xorg.conf file for the SGI Virtu VN200 is listed below for reference. The X server can be modified by: • • Manually editing xorg.conf file Using the /usr/bin/nvidia-xconfig tool After editing the /etc/X11/xorg.conf file, reboot the system so that the changes can take effect.
System Procedures Section "Module" Load Load Load Load Load EndSection "dbe" "extmod" "type1" "freetype" "glx" Section "InputDevice" # generated from default Identifier "Mouse0" Driver "mouse" Option "Protocol" "auto" Option "Device" "/dev/psaux" Option "Emulate3Buttons" "no" Option "ZAxisMapping" "4 5" EndSection Section "InputDevice" # generated from default Identifier "Keyboard0" Driver "kbd" EndSection Section "Monitor" Identifier "Monitor0" VendorName "Unknown" ModelName "Unknown" HorizSync 30.
Starting the X Window System Monitor Section The monitor section contains information that defines the capabilities of your display You can make specific changes to the monitor section of the xorg.conf file by referring to the documentation that came with the monitor. An example monitor section is shown below: Section "Monitor" Identifier "SDM-P234" VendorName "SNY" ModelName "3d0" HorizSync 28-92 VertRefresh 57-85 ModeLine "1920x1200_60n" 154.
System Procedures Device Section The device section defines the configuration parameters for a specific graphics card. There is typically one "Device" section for each graphics card in the system. The device section can make one graphics card with two channels act like two graphics cards using Screen option. A large number of options are available for the setting the device. It is best refer to the NVIDIA README file for specific information about the graphics card configuration parameters.
Starting the X Window System You must also specify either: Option "SecondMonitorHorizSync" "" Option "SecondMonitorVertRefresh" "" or: Option "HorizSync" Option "VertRefresh" "" "" You may also use any of the following options, though they are not required: Option "TwinViewOrientation" Option "ConnectedMonitor" "" "" TwinView Edge-matched Display The TwinView entr
System Procedures TwinView Edge Blending To configure TwinView displays that require edge-blending (refer to Figure 3-4) between the two display channels with a 10% overlap, use the following entries: Option Option "TwinView" "true" # turn on dual channel mode "MetaModes" "1920x1200, 1920x1200 +1728+0" The +1728 in the option line above commands the graphics card to start the second channel at 1728 offset from the lower left hand corner of the first channel (the origin).
Starting the X Window System Passive stereo uses the TwinView clone mode stereo and is not compatible with any other TwinView settings. Note: Stereo is not compatible with the “Composite extension being set”. Please refer to the NVIDIA README file for other stereo options and limitations. Setting up Multiple X Servers on a Single Graphics Card The NVIDIA graphics card can be configured such that each connected display device functions as a separate X server.
System Procedures Then, create two Screen sections, each using one of the Device sections: Section "Screen" Identifier "Screen SG-0" Device "nvidia0" Monitor "Monitor0" DefaultDepth 24 Subsection "Display" Depth 24 Modes "1600x1200" "1024x768" "800x600" "640x480" EndSubsection EndSection Section "Screen" Identifier "Screen SG-1" Device "nvidia1" Monitor "Monitor1" DefaultDepth 24 Subsection "Display" Depth 24 Modes "1600x1200" "1024x768" "800x600" "640x480" EndSubsection EndSection 58 007-5483-001
Starting the X Window System ServerLayout Section The ServerLayout section defines the input/output keyboard/mouse is associated with a screen and where the multiple screens are located relative to one another.
System Procedures Check the NVIDIA Driver To check that the NVIDIA driver is installed correctly, verify that the xorg.conf file contains the following entry under Devices: Section "Device" Identifier Driver VendorName EndSection "Device0" "nvidia" "NVIDIA Corporation" If the driver is not listed as "nvidia" then the xorg.conf file is not using the NVIDIA accelerated drivers. Re-installation of the Graphics Drivers After a clean install of the OS, the .
Troubleshooting Upgrading Drivers SGI provides updated drivers through SupportFolio. If you upgrade drivers directly from the NVIDIA website, remove the .rpm files listed on page 60 first to ensure that there are no conflicts when installing the new drivers.
System Procedures Graphics The following list describes basic troubleshooting tips that you can use to isolate graphics problems. • Each time X starts a log file is generated in /var/log/Xorg.0.log. Common problems that occur after editing the xorg.conf file can be solved by examining this file. • Look for visual artifacts on display device: • Vertical or horizontal stripes • Pixel drop-outs • Missing colors • Run graphics demos to verify performance characteristics of the graphics hardware.
Graphics • • Use the Supermicro xgicfg command from a DOS-bootable thumb-drive image • Use in-band ipmitool from the operating-system prompt Use the ipmitool command or the AOC-SIMSO web interface to communicate with the BMC Caution: The DEFAULTP.EXE command resets all BMC network settings to their default setting. • If the BMC does not respond, you can reset it to default values with the SuperMicro DEFAULTP.
System Procedures Hardware Replacement Procedures Before You Begin Before working with your server product, pay close attention to “Warnings and Cautions” on page 4. Tools and Supplies Needed • • • Phillips (cross head) screwdrivers (#1 bit and #2 bit) Needle-nosed pliers Antistatic wrist strap and conductive foam pad or ESD-safe worksurface Removing the Front Bezel Remove two screws that secure the front bezel to the rack (Figure 3-5) and remove the bezel.
Hardware Replacement Procedures Removing the Graphics Node from the Rack 1. Power off the system using ipmitool or BMC web interface. 2. Disconnect the AC input power cord from the rear panel of system. 3. Disconnect the video and data cables from the rear panel of the system. 4. Remove the front bezel from the system in accordance with the instructions on page 64. 5. Slide the graphics node out of the rack and place it on an ESD-safe worksurface.
System Procedures Removing the Side Cover 1. Remove the graphics node from the rack in accordance with the instructions on page 65. 2. Remove the two screws (Figure 3-7) that secure the side cover to the chassis and remove the side cover.
Hardware Replacement Procedures Replacing the Power Supplies Note: Power supplies can be hot-swapped while the system is in operation in a rack. Have the replacement power supply ready, so that it can be installed quickly without disrupting the airflow through the chassis. 1. If necessary, remove the front bezel from the system in accordance with the instructions on page 64. 2. Move the power supply latch to release the power supply (refer to Figure 3-8). 3.
System Procedures Replacing a Disk Drive Note: Halt the operating system before replacing a disk drive. Disk drives can be hot-plugged while the system is in operation in a rack. Have the replacement disk drive ready, so that it can be installed quickly without disrupting the airflow through the chassis. 1. If necessary, remove the front bezel from the system in accordance with the instructions on page 64. 2. Release the latch on the front of the disk drive carrier (refer to Figure 3-9). 3.
Hardware Replacement Procedures Replacing the Chassis Fans 1. Remove the side cover from the chassis in accordance with the instructions on page 66. 2. Remove two screws (refer to Figure 3-10) that secure the fan assembly to the chassis. 3. Disconnect the fan power cables from the node board. 4. Remove the fan assembly.
System Procedures Note: Step 3 is not required for the NVIDIA Quadro FX 1700 graphics cards. 3. If necessary, remove the three screws that secure the graphics card to the chassis air baffle (refer to Figure 3-11, B). 4. Remove two screws (refer to Figure 3-11, C) that secure the graphics card to the rear panel. 5. Unseat the graphics card from the PCI riser board connector and lift the card out of the chassis (refer to Figure 3-11, D). 6. Place the graphics card on and ESD-safe worksurface.
Hardware Replacement Procedures Replacing the AOC-SIMSO (BMC) Daughter Card The AOC-SIMSO daughter card (BMC) resides in a slot attached to the node board. The rear panel BMC LAN port is connected to the BMC using a short USB cable. 1. Remove the graphics card in accordance with the instructions on page 69. 2. Press the retaining bracket arms apart and gently lift the end of the daughter card that is not in the bracket (refer to Figure 3-12, A). 3.
System Procedures Removing the PCI Riser Board The PCI riser board connects between the graphics card and the node board. 1. Remove the graphics card in accordance with the instructions on page 69. 2. Remove the screws that secure the riser board to the chassis (refer to Figure 3-13). Two options are available: a. Remove the two screws that attach the riser board inside the chassis. b. Remove the two screws that attach to the outside of the chassis. 3.
Hardware Replacement Procedures Removing the Air Baffles The air baffles deflect the airflow from the chassis fans across the node board and the graphics card. 1. Remove the graphics card in accordance with the instructions on page 69. 2. Remove two screws on each side of the chassis (refer to Figure 3-14). 3. Lift the plate out of the chassis.
System Procedures Replacing Node Board Memory DIMMs 1. Remove the graphics card in accordance with the instructions on page 70. 2. Remove the air baffle in accordance with the instructions on page 73, (refer to Figure 3-15, A and B). 3. Press down on the latches to release the DIMM (refer to Figure 3-15, C). 4. Carefully grasp the DIMM by its edge and pull it up and out of the guide rails.
Hardware Replacement Procedures Replacing Node Board Processors ! Caution: When handling the processor, avoid placing direct pressure on the label area of the fan. ! Caution: Do not place the node board on a conductive surface. A conductive surface can damage the BIOS battery and prevent the system from booting. Always place the node board on an ESD-safe worksurface. 1. Remove the graphics card in accordance with the instructions on page 69. 2.
System Procedures 5. Gently lift the load lever to open the load plate. 6. Use your thumb and your index finger to hold the CPU at opposite sides. 7. Carefully lift the CPU straight out of the socket.
Hardware Replacement Procedures Note: If you are removing the processor so that you can return the node board to SGI, install a PnP cap in the socket. Install a Processor Note: A black pick-and-place (PnP) cap is attached to the load plate to protect the CPU socket. 1. Press the load lever down and away from the retention clasp to release the load plate from its locked position.
System Procedures 2. Gently lift the load lever to open the load plate. Load plate released 3. Use your thumb and your index finger to hold the CPU at opposite sides.
Hardware Replacement Procedures 4. Align pin 1 of the CPU (the corner marked with a triangle) with the notched corner of the CPU socket. ! Caution: In step 5, do not drop the CPU on the socket. Do not move the CPU horizontally or vertically, and do not rub the CPU against any surface or any of the contacts which may damage the CPU and/or contacts. 5. Once aligned, carefully lower the CPU straight down into the socket. 6. Find the corner of the CPU that has a semi-circle cutout below a gold dot (CPU key).
System Procedures 8. Use your thumb to gently push the load lever down until it snaps into the retention clasp. Load lever 9. If the CPU is properly installed into the socket, the PnP cap is automatically released from the load plate when the lever locks. ! 80 Caution: Retain the plastic PnP cap for shipping. The node board must be shipped with the PnP cap properly installed to protect the CPU socket. Shipment without the PnP cap properly installed may void the warranty.
Hardware Replacement Procedures 10. Remove the PnP cap. PnP cap released from load plate 11. Repeat these steps to install a second CPU if desired. Install a Heatsink 1. Ensure that you have replaced the thermal interface material (TIM) between the heatsink and the CPU die. Refer to section “Replace the Thermal Interface Material” on page 83. ! Caution: Use caution when you unpack the heat sink so you do not damage the TIM. 2.
System Procedures 4. Install the two remaining screws (screw 3 and screw 4, shown in Figure 3-17); then, gently tighten all four screws.
Hardware Replacement Procedures Remove the Heatsink SGI does not recommend that the CPU or the heatsink be removed from the node board. However, if you do need to remove the heatsink, use the instructions below to prevent damage to the CPU or the CPU socket. 1. Unscrew and remove the heatsink screws from the node board in the sequence as shown in Figure 3-17. ! Caution: Do not use excessive force when loosening the heatsink. Using excessive force can damage the heatsink, CPU, or node board. 2.
System Procedures Removing TIM from a Processor You should always remove the TIM from a processor before you remove it from the node board. ! Warning: Always wear gloves when performing this procedure to avoid direct skin contact with the TIM, which is a potentially toxic material.
Hardware Replacement Procedures Note: If you plan to re-use the heatsink with a replacement processor, you must clean the TIM off of the heatsink. (Refer to “Removing TIM from a Heatsink” on page 87.) 2. Gently push any loose TIM pieces to the center of the processor. 3. Use a dry lint-free cloth to lift the loose TIM pieces off of the processor.
System Procedures 4. Use a dry, lint-free cloth to wipe any remaining TIM pieces off of the processor. 5. Use a different lint-free cloth moistened with isopropyl alcohol (IPA) to remove any remaining TIM from the processor.
Hardware Replacement Procedures Note: Be sure to remove TIM from the gaps between the processor and the load plate. Removing TIM from a Heatsink If you plan re-use the heatsink with a replacement processor, you need to remove the TIM from the heatsink. Warning: Always wear gloves when performing this procedure to avoid direct skin contact with the TIM, which is a potentially toxic material.
System Procedures 1. Remove the heatsink from the socket and place the heatsink on a secure surface with the bottom facing upward. 2. Use a dry, lint-free cloth to wipe any loose TIM pieces off of the heatsink.
Hardware Replacement Procedures 3. Use a different lint-free cloth moistened with isopropyl alcohol (IPA) to remove any remaining TIM from the heatsink.
System Procedures Replacing the Node Board 1. Remove the graphics card in accordance with the instructions on Figure 3-11. 2. Remove the air baffles (refer to Figure 3-18). Air baffle Screws Figure 3-18 Air Baffles 3. Remove the BMC daughter card in accordance with the instructions on page 71. 4. Remove the PCI riser board in accordance with the instructions on page 72. 5. Remove the memory DIMMs in accordance with the instructions on page 74. 6.
Hardware Replacement Procedures Replacing the Control Panel The control panel is located at the front of the system (refer to Figure 3-19). 1. Remove the disk drives or the front top cover to access the cable that connects to the control panel. 2. Disconnect the cable connector that attaches to the control panel cable. 3. Remove the control panel from the chassis.
System Procedures 92 007-5483-001