This site may earn affiliate commissions from the links on this page. Terms of apply.

The writing has been on the wall for this decision ever since Nvidia announced the GTX 1080, only the company has finally made information technology official: Three and four-way SLI rigs won't exist supported with GTX 1080 and 1070 GPUs, salve for "a select few applications."

This news comes courtesy of PC Perspective, who spoke with Nvidia nearly the issue. While Nvidia isn't requiring enthusiasts to acquire an "Enthusiast Fundamental" to enable more two-way SLI, it's also dropping back up from everything only, we imagine, the scattering of popular applications typically used to break world performance records.

Iii-and-iv-mode multi-GPU mostly doesn't work

While the accented die-hard, well-funded enthusiasts may exist disappointed by this move, the fact is, it makes a great deal of sense. Multi-GPU configurations take never scaled well past two video cards, despite improved support for such configurations from Microsoft and a great deal of commuter work from both AMD and Nvidia over the years. PC Perspective and PC Gamer both wrote articles on SLI scaling with the GTX 980 and 980 Ti respectively, and the story is the same with both GPUs — not only do frame times get to hell, only frame rates often increment marginally, if they increment at all.

980TiScalingSLI

Graph by PC Gamer

The problems with multi-GPU evolution and driver development in general were explored very well in a mail service to GameDev.net. The author of the postal service, Promit, writes:

The first lesson is: Nearly every game ships broken. We're talking major AAA titles from vendors who are everyday names in the manufacture. In some cases, we're talking about breathy violations of API rules — ane D3D9 game never even called BeginFrame / EndFrame. Some are mistakes or oversights — ane shipped bad shaders that heavily impacted performance on NV drivers. These things were 24-hour interval to solar day occurrences that went into a bug tracker. Then somebody would get in, observe out what the game screwed upwards, and patch the driver to deal with it. In that location are lots of optional patches already in the commuter that are simply toggled on or off as per-game settings, and then hacks that are more specific to games — up to and including total replacement of the aircraft shaders with custom versions by the driver team. E'er wondered why nigh every major game release is accompanied by a matching driver release from AMD and/or NVIDIA? There you get…

Multi GPU (SLI/CrossfireX) is [expletive deleted] complicated. You cannot begin to conceive of the number of failure cases that are involved until you lot see them in person. I suspect that more than half of the full software attempt within the IHVs is dedicated strictly to making multi-GPU setups work with existing games. (And I don't even know what the hardware side looks like.) If you lot've always tried to independently build an app that uses multi GPU — especially if, god help you, yous tried to practice it in OpenGL — you may have discovered this insane rabbit hole. There is ONE fast path, and it's the narrowest path of all.

When you showtime adding additional complexity in the grade of three or 4 GPUs, you start slamming into multiple system bottlenecks, some of which may or may non be alleviated by DX12. On the plus side, DX12 gives you more than CPU threads to work with, which could help with dispatching piece of work. On the minus side, withal, the new multi-GPU modes crave explicit developer back up (which means the developer is the ane doing the heavy lifting), while the legacy modes may or may non provide the aforementioned scaling capabilities. That'due south earlier nosotros get into technical issues, like PCI Limited bandwidth limitations or the increased overhead required to spin workloads out to more than than two GPUs, merely to put it all back together over again.

Who wrote Pall, Vulkan, and DirectX 12?

While it's unrelated to the residual of this mail, Promit does address a topic that the enthusiast community has argued most for years: Was AMD responsible for the push towards next-generation APIs or not? Promit writes:

So … the subtext that a lot of people aren't calling out explicitly is that this round of new APIs has been done in cooperation with the big engines. The Pall spec is effectively written past Johan Andersson at Dice, and the Khronos Vulkan spec basically pulls Aras P at Unity, Niklas S at Epic, and a couple guys at Valve into the fold…

And so the reason nosotros're seeing new APIs at all stems fundamentally from Andersson hassling the IHVs until AMD woke up, smelled competitive advantage, and started paying attention. That substantially took a three year lag time from when we got hardware to the bespeak that compute could be directly integrated into the core of a render pipeline, which is considered normal today only was bluntly revolutionary at production calibration in 2022. It's a lot of small things adding upwards to a body of water change, with key people pushing on the correct people for the right things.

Just 1 person'south viewpoint, obviously, but an interesting fleck of information notwithstanding. Every bit for three-and-4 way SLI, we uncertainty many will mourn its passing — it's just as well difficult to combine workloads from multiple GPUs without screwing upward frame times and damaging the overall user feel.