What is SLI (Scalable Link Interface)?

SLI is the name of Scalable Link Interface and is simply a method used to connect two or more graphics cards and produce only one output signal.

Generally speaking, SLI is nothing more than a way to connect up to 4 graphics cards, and the system treats them as a single card.

What is SLI in Graphics Card?

The first version of SLI technology was called Scan-Line Interleave. Later, in 1998, they specialized in the production of 3D graphics processors. It was then released by 3dfx Interactive, a company specializing in the design and production of graphics cards that dominated until the late 1990s, especially between 1997 and 2000.

SLI technology became prominent and famous in the Voodoo 2 graphics accelerator chips. Scan-Line Interleave was the first attempt to combine the processing power of two graphics cards. However, they were connected by a small cable that allowed the synchronization information to be shared.

The feature connector was a technology for VGA and SVGA systems that allowed an expansion card to access the main memory of the video card directly without using the system bus.

Nvidia, the company that acquired patents, designers, and technologies from 3dfx, was responsible for launching the technology under the name Scalable Link Interface. In 2004, he modified his work to use new computers with a PCI-Express.

How Does It Work?

Using SLI, it is possible to double the graphics processing power of the computer by adding a second card identical to the first.

You can use two cards from scratch, or you can have one that supports SLI. And when more processing power is needed, you can add a second one. However, there are times when the procedure is more expensive than buying a new graphics card.

The Nvidia implementation requires a motherboard with two PCIe x16 ports. A small printed circuit board connector connects the two cards.

The software distributes the load in two ways. The first analyzes the image to be displayed in a frame, known as SFR (Split Frame Rendering). Thus, it divides the load equally between the two GPUs.

The second form is called AFR (Alternate Frame Rendering). That is, each frame is processed alternately by one GPU. As a result, one frame is processed by the first GPU and the other by the second GPU.

When a box is displayed, the image is sent to the primary GPU over the SLI connection and sent to the output. Ideally, this would cut the rendering time by half, but the actual time would be a bit longer. NVIDIA company says that with this configuration, the system’s performance is increased by 1.9 times.

Types of SLI Technology

NVIDIA has implemented two similar systems that can provide improvements in the graphic performance of a system, up to 2.8x in the case of 3-Way SLI and 1.6x in the case of Quad-SLI, superior to what can be achieved with a simple Graphics card.

Currently, one of the most enjoyable options for creating a cost-effective SLI system, 3-Way SLI, was born with the advent of the GeForce 8 series and, at that time, only the 8800 GTX and Ultra. Today, cards such as the 9800 GTX of the 9 series that have access to this technology are also implemented. These are 3 single-core graphics cards combined with a 6-port connector.

In graphics applications that take advantage of these advantages, as already mentioned, you can see up to 2.8 times higher performance. However, the energy requirements of this configuration can be a significant disadvantage.

For example, in the case of 9800 GTX cards, two additional peripheral connectors are required to provide the power consumption that the simple PCI Express socket cannot offer. And that only 3-way power supplies can provide 1000W.

In response to this problem, NVIDIA launched the Hybrid SLI system, which randomly deactivates the cards that make up the SLI in order to save on parallel consumption. However, this technology is only available for a limited number of motherboards and cards.

Quad-SLI is another implemented SLI option that really leaves much to be desired. Although at first glance, it seems like a simple classic two-card SLI, it is a combination of two cards. In reality, it turns out to be an explosive combination of 4 cores because each card requires the use of only a single PCI Express, combined for technology very similar to SLI inside.

The result is a 4-core SLI system, which can be interpreted as four combined single-core cards and can actually be considered two “two in one.”

The problem is that the very high cost does not justify the performance improvements, which are only 1.6 times greater. This means that the more cores, the lower the total performance multiplication. Also, this improvement can only be seen in a much more limited number of applications than in 3-way. For example, Crysis is a video game with high requirements that only shows the most significant graphical effects when running on this configuration.