The full name of SLI is Scalable Link Interface, also known as Speed Force, which is a patented technology of NVIDIA Company. It uses two PCIE graphics cards of the same model on a motherboard supporting dual PCI Express X 16 through a special interface connection mode.
There are two rendering modes for SLI in NVIDIA: Split Frame Rendering (SFR) and Alternate Frame Rendering (AFR). Splitter frame rendering mode is to divide each frame into upper and lower parts. The main graphics card completes the upper part of the screen, and the auxiliary graphics card completes the lower part. Then the auxiliary graphics card transmits the rendered picture to the main graphics card, and the main graphics card renders with its own upper part. In alternate frame rendering mode, one graphics card is responsible for rendering odd frames and the other graphics card is responsible for rendering even frames. In this mode, the two graphics cards are actually rendered images, and there is no need to connect the main graphics card of the monitor for image synthesis.
In the SLI state, especially in the framing rendering mode, the two graphics cards are not equal. In operation, one graphics card is the master card and the other is the slave card, in which the master card is responsible for the operation and control of task allocation, rendering, post-synthesis, output, etc., while the slave card only receives the tasks from the master card for relevant processing, and then sends the results back to the master card for synthesis and output to the display. Because the main graphics card not only has to complete its own rendering task, but also undertakes the synthesis of the return signal of the auxiliary graphics card, and its workload is much larger than that of the auxiliary graphics card. In addition, in SLI mode, only one monitor can be connected, and multi-head display is not supported.