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

The writing has been on the wall for this conclusion e'er since Nvidia announced the GTX 1080, simply the company has finally made it official: Iii and four-way SLI rigs won't exist supported with GTX 1080 and 1070 GPUs, save for "a select few applications."

This news comes courtesy of PC Perspective, who spoke with Nvidia about the result. While Nvidia isn't requiring enthusiasts to larn an "Enthusiast Key" to enable more than ii-way SLI, it's likewise dropping support from everything but, we imagine, the scattering of pop applications typically used to break world performance records.

3-and-four-way multi-GPU mostly doesn't work

While the absolute die-hard, well-funded enthusiasts may be disappointed by this move, the fact is, it makes a great deal of sense. Multi-GPU configurations have never scaled well past 2 video cards, despite improved support for such configurations from Microsoft and a swell 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 merely practise frame times get to hell, merely frame rates oft increase marginally, if they increment at all.

980TiScalingSLI

Graph past PC Gamer

The problems with multi-GPU evolution and driver evolution in full general were explored very well in a post to GameDev.cyberspace. The author of the mail 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 industry. In some cases, we're talking about blatant violations of API rules — one D3D9 game never fifty-fifty chosen BeginFrame / EndFrame. Some are mistakes or oversights — one shipped bad shaders that heavily impacted functioning on NV drivers. These things were day to day occurrences that went into a issues tracker. Then somebody would get in, find out what the game screwed up, and patch the driver to deal with it. At that place are lots of optional patches already in the driver that are simply toggled on or off as per-game settings, so hacks that are more specific to games — up to and including total replacement of the shipping shaders with custom versions by the commuter team. Ever wondered why almost every major game release is accompanied by a matching commuter release from AMD and/or NVIDIA? There y'all get…

Multi GPU (SLI/CrossfireX) is [expletive deleted] complicated. You lot cannot begin to conceive of the number of failure cases that are involved until you encounter them in person. I doubtable that more half of the total software effort inside the IHVs is dedicated strictly to making multi-GPU setups piece of work with existing games. (And I don't even know what the hardware side looks like.) If you lot've e'er tried to independently build an app that uses multi GPU — especially if, god assist you, you tried to do it in OpenGL — you may have discovered this insane rabbit pigsty. There is One fast path, and information technology's the narrowest path of all.

When you start calculation additional complexity in the form of three or four GPUs, yous outset slamming into multiple organization bottlenecks, some of which may or may not be alleviated by DX12. On the plus side, DX12 gives you more CPU threads to work with, which could assist with dispatching work. On the minus side, however, the new multi-GPU modes crave explicit programmer support (which means the developer is the one doing the heavy lifting), while the legacy modes may or may not provide the same scaling capabilities. That's before nosotros get into technical bug, like PCI Limited bandwidth limitations or the increased overhead required to spin workloads out to more than than 2 GPUs, only to put it all dorsum together again.

Who wrote Mantle, Vulkan, and DirectX 12?

While it's unrelated to the rest of this post, Promit does accost a topic that the enthusiast community has argued almost for years: Was AMD responsible for the button towards next-generation APIs or not? Promit writes:

Then … 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 finer written by Johan Andersson at Die, and the Khronos Vulkan spec basically pulls Aras P at Unity, Niklas S at Epic, and a couple guys at Valve into the fold…

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 twelvemonth 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 but was bluntly revolutionary at production scale in 2012. It's a lot of small things calculation up to a body of water change, with key people pushing on the right people for the right things.

Just one person'south viewpoint, obviously, but an interesting bit of data all the aforementioned. Equally for three-and-four style SLI, we uncertainty many will mourn its passing — it'south just as well difficult to combine workloads from multiple GPUs without screwing up frame times and damaging the overall user experience.