The Onyx 350 InfinitePerformance (IP) graphics and compute module integrates a single InfinitePerformance graphics pipe with a compute module. Each IP graphics pipe in a system can support a single-user display, or the pipes can be connected to an optional compositor. The compositor allows the combined graphics power of two or more IP pipes to be directed to a single display, projector, or other device.
As with the standard Onyx 350 compute module, the IP graphics/compute module supports both PCI and PCI-X protocol, USB keyboard and mouse connections, and optional base or professional audio cards.
The following sections describe the function and physical components of the InfinitePerformance graphics module. All other components on the module function as described in the information on the Onyx 350 compute module. See Chapter 3, “Compute Module”, for more information. Note that Onyx 350 systems always use USB connections as the standard keyboard/mouse connection.
 | Note: Because of the placement of the IP graphics card within the module, the option of a redundant cooling fan or DigitalMediaPro DM3 option is not available with this unit. |
The following sections describe the Onyx 350 InfinitePerformance (IP) compute and graphics module:
The 2U-high compute module with InfinitePerformance (IP) graphics has two free PCI/PCI-X card slots to support an optional audio or other PCI or PCI–X card. The other two PCI slots are occupied by the standard USB keyboard/mouse card and IO9 card.
Note that only the first (referred to as the “base”) IP graphics compute module in your system will have the IO9 PCI interface card, internal serial daughter card, and USB card installed as standard features. You will plug the system USB keyboard/mouse into this module. The optional audio card connections will differ based on the type ordered; an example is shown in Figure 4-1. Note that the connector labeled “Digital out” is the audio output. The figure also shows the location and ports on the IP graphics card, USB card, IO9 interface card, and optional base audio card.
Your Onyx 350 IP graphics module can interconnect to one additional module to make a second IP graphics interface available (a two-pipe system). An optional NUMAlink module is available for interconnecting additional graphics or other optional modules. A maximum of eight IP graphics pipes are supported in the Onyx 350.
Figure 4-2 shows a block diagram of the Onyx 350 with IP graphics.
This section describes the external connectors on the back of the InfinitePerformance graphics pipe supported by your Onyx 350 IP system.
Table 4-1 shows the cable pinout assignments for the graphics board Genlock port.
Table 4-1. Genlock Pinout Assignments
Pin | Assignment |
|---|---|
1 | Genlock input/output video or 3.3V TTL signal levels |
2 | Signal return ground |
Table 4-2 shows the cable pinout assignments for the graphics board Stereo View port.
Table 4-2. Stereo View Pinout Assignments
Pin | Assignment |
|---|---|
1 | +12V DC output to StereoView device |
2 | Ground |
3 | Stereo left/right Eye signal (1=left, 0=right) (STEREO_LEFT) |
Table 4-3 shows the cable pinout assignments for the graphics board Swap-Ready port.
Table 4-3. Swap-Ready Pinout Assignments
Pin | Assignment |
|---|---|
1 | Swapbuffer Gang Sync open collector I/O |
2 | Signal return ground |
Figure 4-6 shows the DVI-I video port.
Table 4-4 shows the port pinout assignments for DVI-I port(s).
Table 4-4. DVI-I Video Port Pinout
Pin | Assignment | Pin | Assignment |
|---|---|---|---|
1 | DATA 2- | 16 | HOT_POWER |
2 | DATA2+ | 17 | DATA 0- |
3 | SHIELD 2/4 | 18 | DATA 0+ |
4 | DATA 4- | 19 | SHIELD 0/5 |
5 | DATA 4+ | 20 | DATA 5- |
6 | DDC_CLOCK | 21 | DATA 5+ |
7 | DDC_DATA | 22 | SHIELD CLOCK |
8 | VSYNC | 23 | CLOCK - |
9 | DATA 1- | 24 | CLOCK + |
10 | DATA1+ | C1 | A_RED |
11 | SHIELD 1/3 | C2 | A_GREEN |
12 | DATA 3- | C3 | A_BLUE |
13 | DATA 3+ | C4 | HYNSC |
14 | DDC_POWER | C5 | A_GROUND2 |
15 | A_GROUND1 | C6 | A_GROUND3 |
Most of the product options you might use in your Onyx 350 system will be either PCI-based or installed in an additional module.
An option that can be ordered with your Onyx 350 IP system without use of an additional module is the Silicon Graphics Dual Channel Display daughterboard (hereinafter referred to as “DCD”) With the DCD, you can expand your viewing area by displaying graphics and text output from one graphics card to two monitors. You can also use the DCD to connect a single monitor. An example of the daughterboard is shown in Figure 4-7.
The DCD connects to the daughterboard slot on the InfinitePerformance graphics board. If you did not order your DCD option from the factory, it must be installed by a trained SGI field service representative on site.
The two DVI-I video/monitor connectors on the DCD option have the same pinouts as those described in Table 4-4.
Before connecting your monitor(s) to the DCD, ensure that you have the proper cables to match your monitor type, either DVI-D (digital) or DVI-A (analog). To connect monitors to the DCD, see Figure 4-8 and use the following steps:
Connect the left monitor's cable to the left video connector (channel 0) on the DCD.
Channel 0 drives the left monitor with the image from the top of the frame buffer.
Connect the right monitor's cable to the right connector (channel 1) on the DCD.
Channel 1 drives the right monitor with the image from the bottom of the frame buffer.
Tighten the thumbscrews on both sides of each video connector.
Note: To ensure proper cursor movement, make sure you connect the monitor cables to the correct video connectors on the rear of your system graphics board.
Connect the power cable for each monitor as follows (see Figure 4-9):
Connect the female end of the power cable to the power connector on the back of the monitor.
Plug the male end into a three-prong grounded electrical outlet.
Plug your system's power cord into an electrical outlet.
Power on your system.
Press the power switch on the left monitor, and then press the power switch on the right monitor to power on the monitors.
For basic information on configuring the DCD, proceed to the next section.
When the X server starts, it checks the user-defined display setting to ensure that it is valid for the current hardware configuration. If a display setting is not specified or if it is invalid, the X server automatically selects the default setting. If you do not want to use the default setting, follow these steps:
| Note: Make sure that both of the monitors that are connected to a single DCD have the same or similar display capabilities. |
Open a UNIX shell.
-
The Graphics Back End Control window appears, as shown in the example in Figure 4-10.
To enable dual channel mode, select a DCD setting from the Valid Formats field. To disable dual channel mode, select a single channel display setting from the Valid Formats field.
Click the Load button next to the Valid Formats title bar.
If your display setting has the character D in the first column, a confirmation dialog box will appear, as shown in the example in Figure 4-11.
Note: If you select a display setting that does not show the character D in the first column, you must log out, and then log in again to activate the settings, as explained later in step 6. Click OK to confirm your display setting.
Another dialog box will appear that asks if you want to use this display setting as the power-on default, as shown in the example in Figure 4-12.
Click OK to use the new display setting as the power-on default, or click Cancel to retain the current power-on default.
If you click OK, the new display setting is immediately activated.
Select the desired frame buffer depth in the Graphics Back End Control window (see Figure 4-13), and then click the Load button.
A dialog box appears that asks if you want to use this frame buffer depth as the power-on default, as shown in the example in Figure 4-13.
Click OK to use this frame buffer depth as the power-on default, or click Cancel to retain the current default.
If you click OK, another dialog box appears that says you must log out for the new settings to take effect, as shown in Figure 4-14.
Click OK in the dialog box.
Select the desired Accumulation Buffer Type in the Graphics Back End Control window (see Figure 4-10), and then click the Load button.
A dialog box appears that asks if you want to use this accumulation buffer type as the power-on default, as shown in the example in Figure 4-15.
Click the OK button to use this accumulation buffer type as the power-on default, or click the Cancel button to retain the current default.
If you click OK, a dialog box appears that says you must log out for the new settings to take effect, as shown earlier in Figure 4-14.
Click OK in the dialog box.
Exit xsetmon and close all active applications.
Log out and then log in again to activate your new settings.
Because the DCD provides a large logical display (for example, 2560 x 1024), some applications use all the available space and display a single window across both of your monitors. If this happens, you can specify the maximum size of a window as follows:
As root, use an editor such as NEdit to open the file /usr/lib/X11/app-defaults/4DWm.
Under 4Dwm Specific Appearance and Behavior Resources, enter the following:
*maximumMaximumSize: 1280x984
This constrains the maximum window size to 1280 x 1024. The 40-pixel vertical difference is for the title bar and the top and bottom window borders.
Save the file and exit the editor.
Restart Window Manager by logging out and logging back in, or by selecting Toolchest > System > Utilities > Restart Window Manager and clicking OK.
The above procedure limits the size of a maximized window, but the entire window may not appear on one of your monitors. To display the window on one of your monitors, click the Maximize button, and then move the window to the desired monitor.
The DCD displays a single logical screen across two monitors. Most applications position the popup windows near the main window, or near the cursor. However, some applications center their popup windows. When such applications are in dual-channel mode, one half of the window appears on one monitor, and the other half of the window appears on the other monitor, as shown in Figure 4-16.
To work around this, modify the application's resources, as follows:
To launch an application in a specific location, add the -geometry option to the command line. For example, the following command opens a window with the upper left-hand corner of the window 30 pixels from the left of the screen and 200 pixels from the top of the screen.
xterm -geometry +30+200
If you are using a resolution of 1280 x 1024 and you want to place a window on the second display, add 1280 to the first number. Note the following example:
xterm -geometry +1310+200
You can also set this X resource in $HOME/.Xdefaults. For example, the following command forces all XWsh windows to open with the upper left-hand corner of the window 30 pixels from the left of the screen and 200 pixels from the top of the screen.
XWsh*geometry: +30+200
XWsh is the application's Classname.
In addition, you can use the Window Setting control panel to set specific window locations or to specify the window's last (continuous) position before you log out. To do this, select Toolchest > Desktop > Customize > Windows.
When you move a window from one monitor to the other, the window follows the cursor as it jumps between screens. However, as it moves across, a section of the window is clipped. For example, as you move a window from the left monitor to the right monitor, the right edge of the window is clipped to the left edge of the right monitor, as shown in Figure 4-17.
