NVIDIA developed a new technology for their video cards called SLI, which stands for Scalable Link Interface. This allows two video cards to be connected or "linked" together by a hardware bridge, allowing them to combine their capabilities on a single monitor.

About SLI

An NVIDIA SLI system includes a PCI Express motherboard that supports two physical connectors that are capable of having two NVIDIA-based PCI Express graphics cards plugged into them. Joined by the NVIDIA SLI connector, the two graphics cards power one monitor, delivering earth-shattering PC performance

Enabling SLI


In order to run SLI mode for your GeForce 7800 GTX 512MB video cards, you must first have both cards installed, connected together with the SLI Connector, and the correct video driver loaded for this configuration. Connect your single display to the top connector on the primary controller (the one closest to the processor.) If everything is correct, you should get the following message whenever you boot your computer:


Enabling LSI mode for your video cards is as simple as checking a box. In the SLI multi-GPU section of the NVIDIA Control Panel, put a check mark in the box labeled "Enable SLI multi-GPU".



With SLI Enabled, a different message will appear indicating you are now in SLI mode, and that your video will only function on your primary monitor:





To disable SLI mode, simply go into the NVIDIA Control Panel and uncheck the "Enable SLI multi-GPU" box, Apply, and reboot.
The SLI high-performance technology combines and scales graphics performance by having multiple NVIDIA video cards (GPUs) in a single system. SLI technology works by intelligently scaling geometry and fill-rate performance for these two GPUs. NVIDIA SLI technology is application-transparent and is enabled through the graphics drivers. All DirectX and OpenGL applications can take advantage of NVIDIA SLI.


In computer graphics, the calculated value of each pixel does not depend on any pixel in another part of the screen. Therefore, the values for pixels in different parts of the screen may be computed in parallel. This is accomplished either through Alternate Frame Rendering, or Split Frame Rendering.



The SLI high-performance technology combines and scales graphics performance by having multiple NVIDIA video cards (GPUs) in a single system. SLI technology works by intelligently scaling geometry and fill-rate performance for these two GPUs. NVIDIA SLI technology is application-transparent and is enabled through the graphics drivers. All DirectX and OpenGL applications can take advantage of NVIDIA SLI.


In computer graphics, the calculated value of each pixel does not depend on any pixel in another part of the screen. Therefore, the values for pixels in different parts of the screen may be computed in parallel. This is accomplished either through Alternate Frame Rendering, or Split Frame Rendering.

The SLI high-performance technology combines and scales graphics performance by having multiple NVIDIA video cards (GPUs) in a single system. SLI technology works by intelligently scaling geometry and fill-rate performance for these two GPUs. NVIDIA SLI technology is application-transparent and is enabled through the graphics drivers. All DirectX and OpenGL applications can take advantage of NVIDIA SLI.


In computer graphics, the calculated value of each pixel does not depend on any pixel in another part of the screen. Therefore, the values for pixels in different parts of the screen may be computed in parallel. This is accomplished either through Alternate Frame Rendering, or Split Frame Rendering.

The main technologies behind NVIDIA SLI are:

Advanced compositing, rendering, and scanout technology, provides uncompromised image quality.
PCI Express bus, which provides bandwidth for maximum system performance.
Innovative inter-GPU communication protocol.
Patent-pending dynamic load-balancing algorithms.

Inter-GPU Communication


The GPUs add to the performance enhancements by having a dedicated connection to transfer digital data. This port is known as the SLI communication port.


The SLI communication port features dedicated scalability logic to facilitate inter-GPU communication. The GPUs communicate with each other directly, so they are not gated by needing to wait for system memory or for internal bus latencies. The SLI communication port transfers image data between the GPUs at a rate of up to 1 GBps, with no PCI Express bus overhead.


NVIDIA SLI Scalability Connector

The interface between the two GPUs on an NVIDIA SLI-enabled system is a high speed digital interface that transfers data at speeds up to 1 GBps.
A variety of data can be transmitted using the scalability connector. This device connects the two video cards.



Dynamic Load-Balancing Software


NVIDIA SLI technology includes software drivers that use dynamic load balancing to share the work between GPUs. The SLI technology drivers control how work is allocated between each GPU.


When the system boots, SLI technology designates one card as master and the other as slave.* The driver, through its patent-pending load-balancing algorithms, determines the load allocation for both cards.


The SLI software drivers select the optimum rendering mode for a given application. The two most basic rendering modes are “alternate frame rendering” and “split frame rendering.”

When SLI technology drivers select alternate frame rendering, one GPU renders a frame and the other GPU renders the next frame. Each GPU sends its frame to the monitor.

Split Frame RenderingSplit Frame RenderingWhen the SLI technology drivers select split frame rendering, each GPU renders a portion of each scene instead of rendering alternate frames. When the system renders a symmetric scene, the load between the two GPUs is 50/50. If a scene is more complex on the top or bottom, the SLI technology driver changes the workload ratio between the two GPUs: one card computes the larger, less complex portion of the scene, and the other GPU computes the smaller, more complex portion of the scene.


When both portions of the image are rendered, the master card uses digital composting to join the two sections and sends the frame to the monitor. Because the images are digital, the SLI technology avoids problems that can arise with analog compositing, such as tearing and artifacts.

Split frame rendering provides high image quality and incredible speed for great performance. The diagram above illustrates how GPUs compute separate portions of the same image.

Load BalancingLoad BalancingIt is possible to see the Load Balancing feature working when the two cards are connected and running in SLI mode. Under the SLI tab, there is an option box that states:

Show GPU Load Balancing
When this box is checked and the user exits Display Properties, they should see a green bar and a green line appear on their display. The green bar indicates the Load Balancing Display Mode is enabled, and the green line delineates where the screen is being "shared" between the two GPUs (Graphic Processing Units). Below are some example screen shots:

This first picture is from 3DMark05. The picture is pretty consistent as to what is being shown on both the upper and bottom halves, so the load-balancing line is midway between the top and bottom. This indicates the two GPUs are equally sharing the screen.

On this second image from the game Far Cry, the top half of the display does not have much to do in the way of activity or motion, nor is it intensive in things like color depths, shadowing, or other difficult graphical presentations. Therefore, the Load Balancing line is shifted more towards the bottom of the screen so the upper GPU can take on additional tasks. This allows the bottom GPU to run faster on the items it is responsible for, which is where most of the activity is taking place.[img][/img][img][/img][img][/img][img][/img][img][/img][img][/img]