Scalable Link Interface

Scalable Link Interface, short SLI, is a multi-GPU technology from NVIDIA that enables the interconnection of two or more graphics chips to improve performance when rendering (SLI Frame Rendering ), or the use of up to four screens ( SLI Multi View).

Theoretically, the rendering performance depending on the operating mode in the middle of the processing power of the GPU's processor cores slowest times the number of GPUs used. When two identical graphics cards thus a theoretical doubling of computing power is achieved in practice, however, performance improvements of about 30 to 90 % are possible - in some applications the computational performance, however, is lower than that of a single graphics card.

History

For the first time appeared the term SLI than Scanline Interleave in 1998 is the previous graphics chip maker 3dfx, the ( Voodoo Graphics, Voodoo 2 and VSA -100) enabled the coupling of several of its graphics chips. Nvidia SLI was only introduced in 2004. Up on the shortcut and the idea of ​​frame under SLI rendering to distribute the computational load on two or more graphics chips, but both methods have nothing in common with each other, although the term SLI was further used after the acquisition of 3dfx by Nvidia.

Before Nvidia SLI published, tried the graphics chip maker ATI already with Alternate Frame Rendering. Today's improved multi -GPU technology from ATI called " Crossfire " and partly differs from Nvidia's SLI technology. The manufacturer also presented XGI 2003 before two graphics cards with multi-GPU technology.

Requirements

Requirement for the use of SLI are firstly a motherboard with a SLI -capable chipset and on the other more than one SLI - capable graphics card, such as onboard graphics or additional PCI - E slots. So far limited the use of NVIDIA SLI largely on their own chipsets, which in addition have been given the code SLI. Only the announcement of Intel Nehalem chipset X58 can be unlocked by the manufacturer on payment of a license fee to NVIDIA for SLI. For this restriction, there are no technical reasons, but it is purely a marketing decision by NVIDIA. There are modified driver which allow the use of SLI on other chipsets. However, these are mostly outdated and there are currently no efforts to unlock SLI - compatible drivers the latest generation, as new drivers seem to regularly and the effort is not worth it.

Initially was to connect the card a special SLI bridge needed ( in German " bridge " ), so that only graphics cards that have these terminals could be connected together. Meanwhile NVIDIA SLI has also allows for cards without this connection, the necessary exchange of data between the cards is handled through the PCI - Express. Since this is limited in bandwidth, this type of SLI is only useful for relatively slow graphics cards, faster should be still connected via the SLI bridge.

In addition, SLI works only with identical graphics chips, partly it even has the exact same graphics card with the same BIOS version be. With each new driver but the SLI feature is flexible. It should also be noted that, worked to driver version 180.xx, upon activation of SLI Frame Rendering only one monitor output. Now, several monitors can be used with the active SLI Frame Rendering.

Mikroruckler

Both Crossfire and SLI is mainly on Alternate Frame Rendering. However, this depends at present often the phenomenon that every second frame is calculated much longer than the previous one. This leads to an uneven flow of the game, which is annoying especially at relatively low frame rates and is often referred to as " micro stuttering ". Due to this fact cards in SLI mode need (including dual-GPU cards) in the respective games significantly higher average frame rates than a single card, to still ensure a relatively smooth representation for the eye.

With the latest drivers the problem of micro stuttering was attenuated. Here, the very fast frame calculated at the output to the screen by an intermediate buffers is slightly delayed. The frame rates are thus again more evenly and the maximum latency between two consecutive frames are smaller, because the waiting time is reduced to the following, slowly calculated frame. With this workaround, take significantly less Microruckler on and the games are better Playable, even if the average FPS number drops slightly

Modes of operation

SLI has the following operating modes:

  • SLI Frame Rendering: Compatibility mode ( compatibility mode). In compatibility mode, only works a card or a graphics chip, a power increase therefore do not exist.
  • Alternate Frame Rendering (AFR )
  • Split Frame Rendering (SFR )
  • Multiview to drive 4 or more monitors. ( Graphics cards work independently in this mode. )
  • Surround technology. When using two or three GPUs in SLI mode more than two monitors (3 monitors) can thereby be controlled. In contrast to the multi-view approach it must not be dispensed with SLI Frame Rendering.
  • 3D Vision Surround. When using two or three GPUs in SLI mode (also for DUAL GPU cards such as GTX 590) three monitors to output stereoscopic image content can be used.

SLI graphics chips

The following graphic chips support SLI:

  • GeForce 6150
  • GeForce 6200 series ( only as slave card )
  • GeForce 6600 series
  • GeForce 6800 series
  • GeForce 7xx0
  • GeForce 8x00 series
  • GeForce 9x00 series
  • GeForce 2xx series
  • GeForce 4xx series
  • GeForce 5xx series
  • GeForce 6xx series
  • GeForce 7xx series
  • Quadro FX 540
  • Quadro FX 1400
  • Quadro FX 3450
  • Quadro FX 4400
  • Quadro FX 4500

Although it takes more graphics cards with the same graphics chip, but not more of the same manufacturer (exception: Quadro FX series).

Dual-GPU graphics cards

In addition, there are graphics cards that are already equipped with two graphics chips and thus are already running in SLI mode.

So you can realize with only one slot for graphics card on the motherboard, a multi-GPU system in two slots Quad -SLI is possible.

On June 6, 2006, before the first NVIDIA multi-GPU card " 7950 GX2 " for the mass market. The "2" at the end of the name refers to the two 7900GT GPUs; this run, however, well below the reference clock of a single 7900GT card because the two existing heat sink for the otherwise waste heat would be too small.

In March 2008, nVidia presented again before a dual-GPU graphics card under the name GeForce 9800 GX2. In the 200 series, the name was discontinued but the dual- GPU card in the series is the Geforce GTX 295 nVidia relied again on the same design principle and built the GPUs on two interconnected boards. Instead of two cooler but this time we built a common condenser, which is located between the two boards. Corresponding recesses in the plates ensure the air supply to the fan.

Quad SLI

If two Nvidia dual- GPU graphics cards are switched together ( exceptions: Asus EN7800GT DUAL, Gigabyte Dual 6600GT ), you get a quad - SLI system (Quad- SLI). However, the performance increase is rather disappointing, especially when considering the high cost of such a configuration. At CeBIT 2006, there were to see first winning performance, stable quad -SLI systems.

3-way SLI

3-way SLI (Triple- SLI) allows the connection of three Nvidia GeForce graphics cards of the same model with 3-way SLI support via a special SLI bridge in a system. The first chipsets that support 3-way SLI, are the nVIDIA nForce 680i SLI, nForce 780i SLI, nForce 780a SLI, nForce 790i SLI and the nForce 790i Ultra SLI.

The following graphics cards support 3-way SLI:

  • Geforce 8800 GTX and GeForce 8800 Ultra
  • GeForce 9800 GTX ( ) and the GTS 250
  • Geforce GTX 260, GTX 275, GTX 280 and GeForce GTX 285
  • Geforce GTX 570, GTX 580
  • Geforce GTX 670, GTX 680
  • Geforce GTX 770, GTX 780
  • Geforce GTX Titan

4- way SLI

In addition to quad -SLI (consisting of two dual-GPU graphics cards), there are also 4-way SLI, but that is not to be confused with Quad SLI. 4-way SLI is realized by four Nvidia GeForce graphics cards of the same model with 4-way SLI support, a 4-way SLI supporting motherboard and a special SLI bridge. The fact that four PCI Express x16 slots the motherboard are needed, but there are not enough PCI Express lanes the chipset available, is often resorted at 4 -way SLI on so-called " bridge chip ". This bridge chips (eg Nvidia NF200 ) help to have enough lanes are available and no power is lost.

The following graphics cards support 4-way SLI:

  • EVGA GeForce GTX 285 Classified
  • Geforce GTX 465, GTX 470, GTX 480 ( Driver 197.75 WHQL from Geforce )
  • Geforce GTX 580
  • Geforce GTX 680
  • Geforce GTX 780
  • Geforce GTX Titan

Hybrid SLI

With Hybrid SLI it is possible to add the waiting, the GPU in the chipset at a low-cost graphics card and then operate in SLI mode. This feature is called by nVidia " GeForce Boost". In addition, you can see the chipset GPU in 3D mode turned off completely (which is necessary with stronger graphics cards, otherwise the chipset GPU would slow down the external graphics card ) and can render only the external graphics card. In the Windows operating is then switched back to the chipset GPU and external graphics card disabled. This process is called NVIDIA "Hybrid Power". First chipsets that support Hybrid SLI, are 730a, 750a SLI and 780a SLI for the AMD platform. However, the graphics card but need to be turned off by the hybrid power control can have an SMBus. The first graphics cards that support Hybrid Power, therefore, the GeForce 9800 GX2 and GeForce 9800 GTX. GeForce Boost is supported by the low-cost graphics cards of the GeForce 8 Series, GeForce 9- series and Nvidia Geforce 9M series. Hybrid Power was only developed for Windows Vista to the beta phase, but was on the grounds that the effort for testing the compatibility of the respective boards with the graphics card is too high and thus not the cost at the rather subdued consumer interest count is adjusted. Since Nvidia retreats further and further out of the chipset business and the two large x86 CPU manufacturer integrating the GPU into the CPU, even the " GeForce Boost" technology will not be developed any further.

A modification of Nvidia's " Hybrid Power " technology and Intel's " Switchable Graphics " is available for mobile graphics chips, the so-called " Optimus " Technogie. This is switched depending on the load back and forth between dedicated GPU from Nvidia and Intel's integrated graphics chip. The commands for changing are doing Nvidia graphics driver, the Nvidia mitgibt a list of 3D applications, based on these switches the driver when starting such an application to the dedicated Nvidia GPU. This feature is only available on Windows 7.

ATI Technologies XGI Technology ATI Multi-Rendering MultiChrome Nvidia ION Tesla (microarchitecture) Nvidia Optimus
404549
de